article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
one of the long - standing experimental and theoretical challenges within nuclear physics is the determination of the nucleus available quantum levels and the decay properties of these levels in the excitation - energy region between the ground state and up to the particle threshold(s ) . in this intermediate excitation - energy region , often called the _ quasicontinuum _ , the _ nuclear level density _ ( nld ) and the average , reduced @xmath0-decay probability , i.e. the _ @xmath0-strength function _ ( @xmath0sf ) , shed light on the dynamic behavior of the nucleus . apart from providing information on basic nuclear properties ,
these quantities are also indispensable input for calculations of , e.g. , neutron - capture cross sections .
these cross sections are of great importance for applications such as the astrophysical heavy - element nucleosynthesis @xcite and modeling of next - generation nuclear power plants @xcite .
amongst a handful of experimental techniques , the oslo method @xcite has been established as one of the promising approaches to obtain experimental information on the nld and @xmath0sf .
the advantage of the oslo method compared to other techniques is that both these quantities can be extracted from one and the same experiment , utilizing typically a charged - particle reaction to record particle-@xmath0 coincidences , in which the structural shape of the nld and the @xmath0sf can be determined . by measuring the energy of the outgoing charged particle ,
the initial excitation energy of the residual nucleus is determined .
the @xmath0 rays de - exciting this initial excitation energy are recorded in coincidence , thus obtaining @xmath0 spectra as function of initial excitation - energy . in 2004 ,
an unexpected enhancement of the @xmath0sf for low transition energies ( @xmath9 mev ) was discovered in the iron isotopes @xmath1fe @xcite .
this feature was not predicted by any theoretically derived @xmath0sfs . in the following years ,
upbend _ was found in many medium - mass nuclei , including @xmath10sc @xcite , @xmath11ni @xcite , @xmath12ge @xcite , and mo isotopes @xcite . to date , the heaviest nuclei where the upbend has been seen are @xmath13la @xcite and @xmath14sm @xcite .
the upbend was experimentally shown to be of dipole nature in @xmath7fe @xcite .
moreover , it has been demonstrated @xcite that such a low - energy enhancement in the @xmath0sf could significantly increase radiative neutron - capture rates of relevance for the @xmath15-process if found to be present in very neutron - rich nuclei . in 2012
, the upbend was independently confirmed in @xmath16mo @xcite using a different technique .
this triggered theoretical investigations of the origin of this phenomenon . within the thermal - continuum quasiparticle random - phase approximation ( tcqrpa ) , the upbend was explained as due to @xmath17 transitions caused by thermal single - quasiparticle excitations in the continuum @xcite , with its strength depending on the nuclear temperature . on the other hand ,
shell - model calculations @xcite show a strong increase in @xmath18 strength for low - energy @xmath19 transitions . at present
, @xmath11ni is the only case where experimental data favor a magnetic character of the upbend @xcite .
more experimental information is needed in order to determine whether the upbend is dominantly of magnetic or electric character , or a mixture of both . in this work ,
we present nlds and @xmath0sfs of @xmath1fe extracted from ( p , p@xmath20@xmath0 ) coincidences .
we analyze systematic errors in the normalization procedure and compare our results to available data in the literature . for the first time
, we present angular distributions of the upbend in @xmath5fe , as well as @xmath0sfs as function of excitation energy to investigate the so - called _ generalized brink - axel hypothesis _ for @xmath1fe .
this hypothesis has up to now only been validated for the heavy nucleus
@xmath21np @xcite .
this article is organized as follows . in section [ sec : exp
] , we give experimental details and the main steps of the oslo - method analysis . in section
[ sec : nldgsf ] , the nlds and @xmath0sfs are shown and the normalization uncertainties are discussed .
further , in section [ sec : ang ] angular distributions are presented for @xmath5fe , while section [ sec : brink ] deals with @xmath0sfs as function of excitation energy and implications for the generalized brink - axel hypothesis .
finally , a summary and outlook are given in section [ sec : sum ] .
the experiments were performed at the oslo cyclotron laboratory ( ocl ) . a 16-mev proton beam with intensity of @xmath22 na impinginged on self - supporting targets of 99.9% enriched @xmath7fe and 92.4% enriched @xmath5fe .
both targets had mass thickness of @xmath23 mg/@xmath24 . accumulating times were @xmath25h and @xmath26h for @xmath1fe , respectively .
the charged ejectiles were measured with the silicon ring particle - detector system ( siri ) @xcite and the @xmath0 rays with the cactus array @xcite .
the siri system consists of eight @xmath27 telescopes .
each telescope is composed of a 130-@xmath28 m thick front detector segmented into eight strips ( angular resolution of @xmath29 ) , and a 1550-@xmath28 m thick back detector . in total
, siri has 64 individual detectors and a solid - angle coverage of @xmath30% . for these experiments ,
siri was placed in forward angles with respect to the beam direction , covering @xmath31 . from the measured energy of the ejectiles and the reaction kinematics ,
the excitation energy of the residual nucleus is deduced .
spectra @xcite ( thick cyan line , scaled with a factor of 70 ) in coincidence with @xmath0 rays measured with the cactus nai detectors for @xmath7fe .
energy bins are 31 kev / channel for protons and 123 kev / channel for @xmath32s .
the first excited levels are marked with their spin / parity . ] in this experiment , the cactus array contained 22 collimated 5 _ in .
_ @xmath33 5 _ in . _
nai(tl ) detectors and six collimated 3.5 _ in .
_ @xmath33 8 _ in .
_ labr@xmath34(ce ) detectors from the milan hector@xmath35 array @xcite .
the nai detectors were placed on the cactus frame with six different angles @xmath36 with respect to the beam direction ( 37.4 , 63.4 , 79.3 , 100.7 , 116.6 , and 142.6 degrees ) , while the labr@xmath34 crystals covered four angles ( 63.4 , 79.3 , 100.7 , and 116.6 degrees ) .
the @xmath0-energy thresholds were @xmath37 kev and @xmath38 kev for the nai and labr@xmath34 detectors , respectively .
particle-@xmath0 coincidences were recorded event - by - event , with the overlap of the @xmath39 and @xmath40 detectors of siri as mastergate for the analog electronics . in total , after background subtraction , about 65 million coincidences were obtained for the nai detectors and about 12 million coincidences for the labr@xmath34 detectors with the @xmath7fe target .
correspondingly , for @xmath5fe , about 15 million and 2.1 million coincidences were recorded for the nai and labr@xmath34 detectors , respectively . in figure
[ fig : particlespectra ] , the proton spectrum of siri in coincidence with @xmath0 rays from the present experiment is compared to the @xmath32 spectrum from the previous experiment reported in ref .
the significant improvement in energy resolution is clear ; the proton spectra have a full width at half maximum ( fwhm ) of @xmath41 kev compared to the @xmath32 spectra where fwhm @xmath42 kev .
the proton-@xmath0 coincidence matrices for the nai and labr@xmath34 detectors are displayed in figure [ fig : matrices ] .
the superior energy resolution for the labr@xmath34 spectra relative to the nai ones is evident , as well as diagonals for which the excitation energy @xmath40 equals the @xmath0 energy @xmath43 corresponding to decay to the ground state .
other diagonals are also clearly visible , for example the direct decay to the first - excited 2@xmath44 state in @xmath7fe .
-ray energy versus excitation energy before unfolding for ( a ) @xmath7fe , nai detectors ; ( b ) @xmath7fe , labr@xmath8 detectors ; ( c ) @xmath5fe , nai detectors ; ( d ) @xmath5fe , labr@xmath8 detectors .
energy bins are 14 kev / channel . ]
it is also very interesting to note the `` triangles '' in the @xmath5fe matrix where the @xmath0 intensity suddenly drops , see for example at @xmath45 mev in figure [ fig : matrices]c , d .
one would naively think that the @xmath0 intensity would be significantly reduced as soon as the neutron separation energy @xmath46 is reached ; however , this is well above @xmath47 mev .
this feature is explained by considering the average spin @xmath48 populated at high excitation energies . from @xmath0 transitions in coincindence with protons
, we identify the decay from the @xmath49 level at @xmath50 mev in @xmath7fe as well as other levels with spins 2 , 3 , 4 , 5 @xcite . levels with these spins will be hindered in decaying through @xmath51-wave neutron emission to the @xmath52 ground state in @xmath7fe .
this hindrance is studied in detail for @xmath16mo and applied in a novel technique to determine spins in @xcite . in order to obtain the correct @xmath0-energy distribution for each excitation - energy bin ,
the signals from the nai and labr@xmath34 detectors must be corrected for the detector response .
we applied the unfolding technique described in @xcite , which is an iterative procedure using a strong smoothing of the compton part of the spectrum . in order to construct response functions for the nai and labr@xmath34 detectors
, we used in - beam measured transitions from @xmath7fe , @xmath53si , @xmath54c , and @xmath55o @xcite .
moreover , we made use of a subtraction technique @xcite to extract the distribution of primary @xmath0 rays for each excitation - energy bin .
this distribution contains information on the nld and the @xmath0sf as deduced from fermi s golden rule @xcite : @xmath56 where @xmath57 is the decay rate between initial state @xmath58 and final state @xmath59 , @xmath60 is the transition operator and @xmath61 is the density of final states .
similarly , the distribution of primary @xmath0 rays as function of @xmath40 depends on the level density at @xmath62 and the @xmath0-transmission coefficient @xmath63 for the @xmath0 transition with energy @xmath43 .
the @xmath0-transmission coefficient is directly proportional to the @xmath0sf .
our ansatz is @xcite : @xmath64 where @xmath65 is the matrix of primary @xmath0 rays , representing relative intensities or branching ratios for a given transition energy @xmath43 at a given initial excitation energy @xmath40 .
the primary @xmath0-ray matrices @xmath65 for @xmath1fe are shown in figure [ fig : fg_rsg_matrices ] .
they are normalized for each excitation - energy bin so that @xmath66 .
this means that the probability for @xmath0 decay from a given bin is 1 , and that the intensity of a given @xmath0-ray energy reflects the branching ratio for that particular transition energy .
these matrices are used as input for the extraction of the nld and @xmath0sf for the four data sets . the expression in equation
[ eq : brink ] is valid for statistical decay , i.e. where the decay is independent of the formation of the compound state @xcite .
this is fulfilled at rather high excitation energies where the initial nld is high , typically above @xmath67 where the pair - gap parameter @xmath68 @xcite .
note that @xmath63 is a function only of @xmath43 and not @xmath40 or @xmath69 , in accordance with the generalized brink - axel hypothesis @xcite .
this will be discussed in detail in section [ sec : brink ] .
rays energy versus excitation energy for ( a ) @xmath7fe , nai detectors ; ( b ) @xmath7fe , labr@xmath8 detectors ; ( c ) @xmath5fe , nai detectors ; ( d ) @xmath5fe , labr@xmath8 detectors . energy bins are 124 kev / channel for @xmath7fe and 120 kev / channel for @xmath5fe .
note the different energy scales for the lower and upper panels . ]
the functional form of the nld and @xmath0sf is determined through a least-@xmath70 fit to the @xmath65 matrices as described in @xcite .
the 3d landscapes as shown in figure [ fig : fg_rsg_matrices ] are used simultaneously in the fit .
the sum of all primary transitions for each @xmath40 bin is normalized to unity .
as the @xmath65 matrices contain many more data points ( `` pixels '' ) than the free parameters ( the vector elements of @xmath71 and @xmath72 ) , the solution is uniquely determined and the fit routine converges fast , typically within 10 - 20 iterations . some considerations need to be made before extracting the nld and @xmath0sf from the data .
first , a low - energy limit for the excitation energy is applied to avoid the discrete region at low @xmath40 , for which the condition of a compound - nucleus decay is highly questionable .
further , an upper limit @xmath73 must be given , which typically corresponds to @xmath46 , as neutrons are not measured or discriminated in the present experimental setup .
finally , a low - energy limit on the @xmath0 energy , @xmath74 , is determined to exclude eventual higher - generation transitions not properly subtracted in the primary - distribution extraction , as discussed in detail in @xcite .
the chosen energy limits for the extraction procedure are : @xmath75 mev , @xmath76 mev , and @xmath77 mev for @xmath7fe ; correspondingly , @xmath78 mev , @xmath79 mev , and @xmath80 mev for @xmath5fe .
the neutron separation energies @xmath46 are 11.197 mev and 7.646 mev for @xmath1fe , respectively .
the reason why we are able to put @xmath73 higher than @xmath46 in the case of @xmath5fe , is that the first - excited state in @xmath7fe is at 847 kev , allowing in principle for @xmath81 mev = 8.5 mev as we are requiring proton-@xmath0 coincidences .
similarly , for @xmath5fe , the upper limit is @xmath82 100 kev above @xmath46 . to test the quality of the fit , which is based on all primary spectra included in the extraction procedure ,
we take the obtained @xmath71 and @xmath72 functions and use them to generate primary @xmath0 spectra to be compared with the input spectra bin by bin .
this is shown in figure [ fig : diw_56fe_nai ] .
error bars in the primary spectra reflect statistical uncertainties , and systematic uncertainties stemming from the unfolding procedure and the extraction of the primary @xmath0 rays @xcite .
spectra for @xmath7fe ( black points , nai detectors ) with the calculated ones ( blue histogram ) from the extracted @xmath83 and @xmath63 functions for a set of initial excitation - energy bins as indicated in the panels .
energy bins are 124 kev / channel . ] for @xmath7fe , using data from the labr@xmath34 detectors . ] for @xmath5fe measured with nai detectors .
energy bins are 120 kev / channel . ] for @xmath5fe using data from the labr@xmath34 detectors . ] as can be seen from figures [ fig : diw_56fe_nai][fig : diw_57fe_labr3 ] , the overall agreement between the data and the calculated primary spectra is very good . it should be noted that porter - thomas fluctuations @xcite of the decay strengths are not taken into account .
these fluctuations are expected to be large when the final level density @xmath61 is low .
this is clearly visible _
e.g. _ in the decay to the first - excited level in @xmath7fe , see figure [ fig : diw_56fe_nai]a and the peak at @xmath84 mev , where data points are several standard deviations off the calculated @xmath85 . here
, there is only one final level and the relative decay strength is seen to fluctuate strongly for different initial excitation energies .
as only the functional form of the nld and @xmath0sf can be deduced from the primary @xmath0 spectra , the slope and absolute normalization must be determined from auxiliary data .
it is shown in @xcite that any solution @xmath86 and @xmath87 will give an equally good @xmath70 fit to the primary-@xmath0 data through the transformations @xmath88\,\tilde{\rho}(e - e_\gamma ) , \label{eq : array1}\\ { \mathcal{t}}(e_\gamma)&=&\mathcal{b}\exp(\alpha e_\gamma)\tilde{{\mathcal{t } } } ( e_\gamma ) , \label{eq : array2}\end{aligned}\ ] ] where the parameters @xmath89 , @xmath90 , are the absolute normalization of the nld and the @xmath0-transmission coefficient , respectively , and @xmath32 is the common slope parameter . for the nld ,
the parameters @xmath89 and @xmath32 are found by fitting our data to known levels from the literature @xcite at low excitation energy and to neutron - resonance spacing data from @xcite at @xmath46 .
the discrete levels are binned with the same bin width as our experimental data . for @xmath7fe , there is no information from neutron - resonance experiments as @xmath91fe is unstable . for this case
, we have estimated the nld at @xmath46 from systematics in the following way : * to estimate the lower - limit nld , we calculate the total level density from the @xmath51-wave neutron resonance spacing @xmath92 for fe isotopes where this value is available from @xcite according to the expression @xmath93 + j_t\exp\left[-j_t^2/2\sigma^2\right ] } , \label{eq : rhosn}\ ] ] assuming equally many positive- and negative - parity states . here
, @xmath94 is the ground - state spin of the target nucleus in the neutron - resonance experiment and @xmath95 is the spin cutoff parameter .
we make use of the phenomenological spin cutoff parameter suggested in @xcite : @xmath96 here , @xmath97 is the mass number and @xmath98 is the deuteron pairing energy as defined in @xcite .
this approach gives a low value for the spin cutoff parameter and thus a low limit for the level density .
further , we calculate @xmath99 from the global systematics @xcite directly . by taking the @xmath70 fit of the semi - experimental @xmath99 with the values from systematics in the same fashion
as done for @xmath100y in @xcite , one obtains an estimate for the @xmath7fe @xmath101 .
all parameters are given in table [ table1 ] .
this normalization is referred to as _
norm-1 _ in the folowing . * to estimate the upper - limit nld , we apply the same procedure as in ( _ i _ ) but with the spin cutoff parameter given by the rigid - body moment of inertia approach as parameterized in @xcite : @xmath102 here , @xmath103 is the level - density parameter and @xmath104 is the excitation - energy backshift determined from global systematics of @xcite .
all parameters are given in table [ table2 ] .
we refer to this normalization as _ norm-2_. for @xmath5fe , we use the @xmath92 value given in @xcite and estimate @xmath99 using equation [ eq : rhosn ] , again with spin cutoff parameters both from @xcite and @xcite .
consistent with the approach for @xmath7fe , the lower limit is obtained with the spin cutoff parameter in equation [ eq : spincut ] , and the upper limit with the one in equation [ eq : spincut05 ] , also including the uncertainties in @xmath92 .
all parameters are listed in table [ table1 ] and [ table2 ] .
[ cols="<,^,^,<,^ , < ,
< , < , < " , ] + [ tab : coeff ] to further study the angular - distribution coefficients for the upbend in @xmath5fe in more detail , we make individual fits of equation [ eq : legendre ] to eight 300-kev wide excitation - energy cuts in the primary @xmath0-ray matrix in the range @xmath105 mev , @xmath106 mev .
the resulting @xmath107 and @xmath108 coefficients are shown in figure [ fig : angcoeff ] .
we obtain @xmath109 and @xmath110 , in excellent agreement with the simultaneous fit to the whole region as shown in figure [ fig : angular_trans]d .
hence , we conclude that the upbend structure in @xmath5fe is also very likely to be caused by dipole transitions , but for this case the non - stretched transitions seem to dominate . and @xmath108 coefficients from independent fits of 300-kev excitation - energy cuts in the @xmath5fe primary matrices for the six cactus angles .
as the labr@xmath8 detectors have excellent energy resolution and efficiency for high - energy @xmath0 rays , we make use of the technique described in @xcite to extract the @xmath0sf as function of excitation energy . we start with the primary @xmath0-ray matrix @xmath111 obtained in section [ sec : exp ] .
we will now make the assumption that the nld is the one determined in section [ sec : nldgsf ] , but the transmission coefficient @xmath112 is now allowed to be dependent on both excitation energy and @xmath0-ray energy@xmath113 . as @xmath71 is known , we can in principle determine @xmath114 for each excitation - energy bin just by dividing the primary @xmath0 matrix with the nld : @xmath115 , using our ansatz in equation [ eq : brink ] . specifically , we have @xmath116 where @xmath117 is a normalization factor in units mev@xmath118 , depending only on the initial excitation energy .
sfs as function of initial excitation energy for @xmath7fe .
bins are 248 kev / channel for @xmath40 and 124 kev / channel for @xmath43 . ] now , this game can be played in two ways : * we investigate @xmath112 as function of _ initial _ excitation energy through the relation @xmath119 we determine @xmath117 by @xmath120 note that @xmath121 is the normalized transmission coefficient from section [ sec : nldgsf ]
. however , it will not influence the _ shape _ of the extracted @xmath122 as it acts as a constant after integrating over all @xmath43 .
hence , it only serves to provide an approximate absolute normalization of @xmath122 .
* we can also find @xmath112 as function of _ final _ excitation energy by @xmath123 where we keep in mind that @xmath124 .
again , we assume that @xmath121 gives a good estimate of the absolute value and we can approximate the normalization for a given final excitation energy @xmath69 and for a specific @xmath43 fulfilling @xmath125 by @xmath126 the @xmath0sf as function of excitation energy is then easily calculated from the transmission coefficient by use of equation [ eq : dipolestrength ] .
the results are shown for @xmath1fe in figures [ fig : radex56fe ] and [ fig : radex57fe ] , respectively .
sfs as function of initial excitation energy for @xmath5fe .
bins are 480 kev / channel for @xmath40 and 120 kev / channel for @xmath43 . ]
we observe that the decay strength to the ground state increases as function of both @xmath40 and @xmath43 , which is fully consistent with the @xmath0sf determined previously in section [ sec : nldgsf ] and the expected influence of the tail from the gdr .
moreover , we find that the @xmath0sf varies with initial excitation energy , but that the general shape is preserved : there is always an upbend at low @xmath43 and a rather flat distribution of strength in the middle @xmath43 region , before it again increases for high @xmath43 .
to investigate the fluctuations , following @xcite , we compare the average @xmath0sf for all initial excitation energies with the @xmath0sf obtained for a specific excitation - energy bin .
we find that the fluctuations relative to the average @xmath0sf can be large , more than 100% for some @xmath0-ray energies and @xmath40 .
also , the fluctuations are in some cases significantly larger than the error bars .
therefore , it seems that although the overall shape of the @xmath0sf is indeed preserved in agreement with the generalized brink - axel hypothesis , the @xmath0sf for a specific transition energy and excitation energy could have a large deviation , in particular when the excitation - energy bin is narrow and containing rather few levels .
sf for the @xmath7fe ground state as @xmath69 ( black points ) and for @xmath127 mev . ]
sf for the @xmath5fe ground - state band , @xmath128 mev ( black points ) and for @xmath129 mev .
] finally , we also investigate the @xmath0sf for a specific final excitation energy .
we have chosen the ground state in @xmath7fe and the ground - state band ( 1/2@xmath130 , 3/2@xmath130 , 5/2@xmath130 ) in @xmath5fe .
the @xmath0sf for this @xmath69 is then compared to a typical @xmath0sf at a high initial @xmath40 , see figures [ fig : rsf_final_56fe ] and [ fig : rsf_final_57fe ] .
again , we observe that the general trend is preserved , although significant deviations are present , for example for the @xmath7fe strength at @xmath131 mev .
this is interpreted to be caused by porter - thomas fluctuations , which are expected to be large when the final and/or the initial nld is low @xcite .
in this work , we have presented data on @xmath1fe from ( @xmath132 reactions , measuring the @xmath0 rays both with nai and labr@xmath34 crystals . from the sets of particle-@xmath0 coincidences , we have extracted nlds and @xmath0sfs for these nuclei .
we find that our present results compare well with previous measurements .
we have measured angular distributions for @xmath5fe , and we get coefficients in agreement with a dominance of non - stretched dipoles for the upbend , in agreement with recent shell - model calculations .
moreover , we have investigated the excitation - energy dependence of the @xmath0sf .
the data show that the general trends are preserved , i.e. the upbend and an enhanced decay strength for high - energy transitions , in accordance with the brink - axel hypothesis .
however , we also encounter large fluctuations , which seem to be due to strong porter - thomas fluctuations caused by the low level density in these light nuclei .
the authors wish to thank j.c .
mller , e.a .
olsen , a. semchenkov and j. c. wikne at the oslo cyclotron laboratory for providing excellent experimental conditions .
this work was financed by the research council of norway ( nfr ) , project grant no . 205528 , and through erc - stg-2014 under grant ggreement no .
s. s. and g. m. t. acknowledge financial support by the nfr under project grant no .
210007 and 222287 , respectively .
a. v. v. acknowledges funding from the department of energy , grant no .
de - na0002905 .
m. w. acknowledges support by the national research foundation of south africa under grant no .
99 arnould m and goriely s 2003 _ phys . rep . _ * 384 * 1 arnould m , goriely s , and takahashi k 2007 _ phys .
rep . _ * 450 * 97 aliberti g , palmiotti g , salvatores m , and stenberg c g 2004 _ nucl .
eng . _ * 146 * 13 aliberti g _ et al .
_ 2006 _ ann .
nucl . energy _ * 33 * 13 schiller a , bergholt l , guttormsen m , melby e , rekstad j , and siem s 2000 _ nucl .
methods phys .
res . a _ * 447 * 498 voinov a _
_ 2004 _ phys .
lett . _ * 93 * 142504 larsen a c _ et al .
_ 2007 _ phys .
c _ * 76 * 044303 brger a _ et al .
_ 2012 _ phys .
c _ * 85 * 064328 voinov a v _ et al .
_ 2010 _ phys . rev .
c _ * 81 * 024319 renstrm t _
_ 2016 _ phys . rev .
c _ * 93 * 064302 guttormsen m _
_ 2005 _ phys .
c _ * 71 * 044307 wiedeking m _
_ 2012 _ phys .
lett . _ * 108 * 162503 tveten g m _ et al .
_ 2016 _ phys .
c _ * 94 * 025804 kheswa b v _ et al .
_ 2015 _ phys .
b _ * 744 * 268 simon a _ et al .
_ 2016 _ phys .
c _ * 93 * 034303 larsen a c _ et al .
_ 2013 _ phys .
* 111 * 242504 larsen a c and goriely s 2010 _ phys .
c _ * 82 * 014318 litvinova e and belov n 2013 _ phys . rev .
c _ * 88 * 031302(r ) schwengner r , frauendorf s , and larsen a c 2013 _ phys .
* 111 * 232504 brown b a and larsen a c 2014 _ phys .
_ * 113 * 252502 guttormsen m _
_ 2016 _ phys . rev .
lett . _ * 116 * 012502 guttormsen m , brger a , hansen t e , and lietaer n 2011 _ nucl . instrum .
methods phys .
res . a _ * 648 * 168 guttormsen m _
_ 1990 _ phys .
t _ * 3254 nicolini r _ et al .
_ 2007 _ nucl .
methods phys .
res . a _ * 582 * 554 giaz a _ et al _ 2013 _ nucl .
methods phys .
res . a _ * 729 * 910 wiedeking m _ et al _ 2016 _ phys
. rev .
c _ * 93 * 024303 guttormsen m _
_ 1996 _ nucl . instrum .
methods phys .
res . a _ * 374 * 371 crespo campo l _ et al .
_ 2016 _ phys .
( in press ) guttormsen m , ramsy t , and rekstad j 1987 _ nucl . instrum
. methods phys .
a _ _ 255 _ 518 dirac p a m 1927 _ proc .
london a _ * 114 * 243 fermi e 1950 _ nuclear physics _
( university of chicago press , chicago ) brink d m 1955 _ dr . phil .
thesis , oxford university _
axel p 1962 _ phys .
_ * 126 * 671 bohr a and mottelson b 1969 _ nuclear structure _ ( benjamin , new york , vol .
i ) larsen a c _ et al .
_ 2011 _ phys .
c _ * 83 * 034315 porter c e and thomas r g 1956 _ phys
. rev . _ * 104 * 483 data from the nndc on - line data service database as of august 2016 ; available at http://www.nndc.bnl.gov/nudat2/ capote r _
_ 2009 _ nucl .
data sheets _ * 110 * 3107 ; _ reference input parameter library ( ripl-3 ) _ , available at http://www-nds.iaea.org/ripl-3/ larsen a c _ et al .
_ 2016 _ phys .
c _ * 93 * 045810 von egidy t and bucurescu d 2009 _ phys .
c _ * 80 * 054310 von egidy t and bucurescu d 2005 _ phys .
c _ * 72 * 044311 ; 2006 _ phys .
c _ * 73 * 049901(e ) ericson t 1959 _ nucl .
_ * 11 * 481 ericson t 1960 _ adv .
* 9 * 425 voinov a , guttormsen m , melby e , rekstad j , schiller a , and siem s 2001 _ phys . rev .
c _ * 63 * 044313 kopecky j and uhl m 1990 _ phys .
c _ * 41 * 1941 voinov a _ et al .
_ 2006 _ phys . rev .
c _ * 74 * 014314 vonach h k _
1966 _ phys .
rev . _ * 149 * 844 voinov a _
_ 2007 _ phys .
c _ * 76 * 044602 fischer f f et al , phys .
c 30 , 72 ( 1984 ) alvarez r a , berman b l , faul d d , lewis jr f h , and meyer p 1979 _ phys .
c _ * 20 * 128 bartholomew g a , earle e d , ferguson a j , knowles j w , and lone m a 1972 _ adv .
* 7 * 229 litherland a e and a. j. ferguson a j 1961 _ can .
_ * 39 * 788 der mateosian e and sunyar a w 1974 _ at . data and nucl
. data tables _ * 13 * 391 krane k s 1978 _ at .
data and nucl .
data tables _ * 22 * 269 data from the evaluated nuclear structure data file ( ensdf ) as of august 2016 , available at http://www.nndc.bnl.gov/ensdf/ guttormsen m _
_ 2011 _ phys .
c _ * 83 * 014312 *
|
nuclear level densities and @xmath0-ray strength functions of @xmath1fe have been extracted from proton-@xmath0 coincidences .
the level densities are compared to previous experimental data , both from particle - evaporation spectra and @xmath2he - induced reactions .
the low - energy enhancement in the @xmath0-ray strength functions , first discovered in ( @xmath3he,@xmath3he@xmath4)@xmath5fe and ( @xmath3he,@xmath6)@xmath7fe data , is confirmed with the experiments reported here .
angular distributions of the low - energy enhancement in @xmath5fe indicate its dipole nature . the high statistics and the excellent energy resolution of the large - volume labr@xmath8(ce ) detectors allowed for a thorough analysis of the @xmath0-ray strength function as a function of excitation energy .
strong fluctuations were found for the direct decay to the ground band , while the decay into the quasi - continuum shows a more uniform behavior .
there is no indication of any significant excitation - energy dependence in the @xmath0-ray strength function , in support of the generalized brink - axel hypothesis .
october 2016 _ keywords _ : level density , @xmath0-strength function , angular distributions , brink hypothesis , iron +
|
hawking set out early on to outline an information loss model@xcite , which drew some significant criticism@xcite .
the full consequences of information loss scenarios are still unknown .
it is easy for non - unitary theories to lose causality , though energy conservation can still be maintained with care@xcite .
this category is characterized by a many - to - one map of states , as well as an injective map from pure to mixed states .
the no - hair theorem is embodied in the first property , in which a nonunitary quantum bleaching operator is applied .
the second property is shared by some other theories of fundamental decoherence@xcite .
+ while energy conservation can often be preserved by careful choice of the decohering basis , the bleaching property introduces strong nonlocality by breaking entanglement with a local process .
if one of a pair of entangled particles is sent into such a nonunitary process , it can be detected by local measurements on the other .
this will generically lead to loss of causality .
a similar consequence follows from violation of the no cloning theorem as well as in many nonlinear generalizations of qm@xcite .
the difference in observable consequences of information loss and hidden sector models can be quite strong .
these models are most characterized by fundamental deformation of quantum mechanics where information is unrecoverable in principle .
+ the `` trace over '' model is what most conservative proponents of information loss refer to as information loss .
it is to simply take the trace over all observables on some boundary inside the horizon , possibly just surrounding the singularity . in this way
each singularity in spacetime represents a sealed box of state information forever unavailable to regular observers .
the accumulation of entropy comes from the gradual accumulation of entanglement with these un - observables. as such , this scenario should belong in the hidden sector class of theories .
one could argue that the thermofield double model is simply a convenient way to group all of these hidden pieces together .
in unitary quantum mechanics any bleaching process moves the entanglement to somewhere else .
hidden sector models put all in falling qubits into some inaccessible region .
stable remnants and baby universes would be examples of this type .
+ maldacena s thermofield double model falls into this type with the hidden sector here being highly nonlocal modes involving entanglement between left and right copies of the boundary field theories@xcite .
there are two troubling issues here .
first , the thermofield double treatment can be used to model virtually any mundane lossy process , where information escapes into a thermal bath . as such
, the duality owes nothing in particular to the specific physics of black holes .
secondly , the image of left and right copies of the matter fields living on two sides of an einstein - rosen bridge does not seem to be physical . in a semi - classical model put onto the maximally extended metric , we simply have information being lost on both sides of the non - traversable wormhole . while matter may become entangled between one asymptotic region and another , the combined state is also still entangled with the singularity .
this is sometimes combined with postselection so that all information ends up exchanged with the bath , leaving none on the singularity .
+ all models of this type predict roughly the same physics , in which hawking radiation is uncorrelated with infalling matter .
entropy , at least as measured by local observers on our side of the black hole , increases monotonically for the entire evaporation process . while the overall state is pure , entanglement entropy is maximized with the second set of fields .
+ combining postselection with the field doubling model gives another interesting theory , with it s own issues . if one takes the full extended metric with second asymptotic region , and then imposes a final state on the singularity , we can time order the interior so that information that would have been lost is ` reflected ' into the past of the second asymptotic region .
this could also be seen as the second region acting as a unitary eraser on the infalling matter states .
this scenario also has two problems .
first it does not get the infalling states into the hawking radiation .
both infalling matter and hawking radiation are erased by their ` mirror image ' from the second initial surface .
this second initial surface acts as an arbitrarily large one time pad , like the thermal bath of thermofield dynamics , resulting in outgoing hawking radiation that is uncorrelated with anything else from our visible asymptotic region .
the dynamics observable from our asymptotic region would then be the same as any other generic hidden sector / loss model .
secondly , it requires that matter falling into the black hole from our side be already entangled with matter in this bath , even before the black hole forms .
this may represent an unacceptable level of fine tuning of the initial state of the ` second copy ' fields , while formally satisfying the unitary dynamics of ads / cft .
+ a second paradox appears in the case of the thermofield model .
suppose the black hole is surrounded by a mirror .
this reflects all the radiation from our region back inward , but has no effect on radiation emitted into the second asymptotic region .
the hole in the second region will evaporate but the one in the first region will not . if the second region is a hidden sector , then both holes should vanish , taking the mass energy away into the hidden sector . on the other hand , if we started with a wormhole , the remaining mouth could be thought of as uncoupled or as a traversable time machine , depending on the completion of the interior .
this is explored by krasnikov@xcite .
there have been a few proposals of this type@xcite .
the black hole final state hypothesis is one , although a second could be constructed out of hawking s cryptic comments about the euclidean action suppressing black hole formation@xcite . in the first case we save information from hitting the singularity by using post - selection to place a final state boundary condition around the singularity .
this boundary condition restricts the formation of hawking radiation to only produce the negative energy modes that complement infalling matter so that the fields take on the correct values on a particular final time slice .
problems with this approach have been brought up@xcite , and it appears to lead to loss of causality .
+ while the reverse causation aspect has been used as a critique of these models , it has the advantage of making a possibly testable prediction . in a manner reminiscent of early big crunch scenarios ,
a generic post - selected field has a unique value on a final surface , and entropy increases on successive spatial slices at earlier rather than later times .
the generic case propagates backwards from the singularity , effectively choosing the advanced rather than retarded modes for infalling fields .
the horizon acts as the turnover surface where retarded modes again begin dominating .
if the turnover is sudden , a firewall may appear at the junction between the two regions . if it is gradual , or if smoothness of the junction is imposed , the transition would presumably be blurred over the duration of the collapse process , allowing some advanced modes to escape to past infinity .
this escaped radiation might be observable using weak measurement techniques , as a collapsing shell of red - shifted radiation that converges during black hole formation .
the only advanced modes that would be excited far from the hole would be those eventually reaching the horizon .
a normal dissipative measurement apparatus or absorber would appear to emit a contribution to the collapsing wave front rather than register a normal detection event .
this is why weak or indirect measurement is needed . for a description of weak measurement and postselection physics , see@xcite .
+ the effect can also be understood to arise from the near saturation of the bekenstien bound in the final moments of collapse .
the time reversed movie of a system s evolution does not display much of an arrow of time if entropy is at or near maximum .
a ten minute movie of mixing a drink may be indistinguishable from the reverse for nine out of the ten minutes .
+ the other aspect of these models is the potential propagation of the postselected weighting factors onto the pre - collapsed matter .
the hawing radiation acts as a thermal bath whose entropy is determined by the complement of the infalling state . in this way postselection may create a boltzmann - like weighting factor for collapsing states , where the relative entropy between the collapsing star and final surface plays the role of a chemical potential .
+ just as post - selection deforms unitary quantum mechanics to create hawking radiation correlations , a more extreme deformation may be used to prevent black holes from forming at all.@xcite if a boundary condition is placed at the horizon we have a picture in which there is no spacetime interior at all , but a mirror - like firewall .
the need for fields to know about global conditions in order to form such structures is satisfied by the nonlocal nature of post - selected systems . in each of these scenarios
it is not clear how to treat more complex quantum states such as superpositions of black holes and regular spacetimes .
the nonlinear nature of post - selection may lead to loss of unitarity anyway for these states .
another approach is to explicitly add nonlocal or even acausal fields to the black hole interior to allow information to escape .
nonviolent nonlocality(nvnl ) attempts to describe how a generic tachyonic field could leak the information while being strongly suppressed at large distances from the black hole@xcite .
whether this effective field arises from a more fundamental structure is another question . +
effective source terms are chosen to comply with a no drama principle at least to first order , to give a sufficient channel for enough information to escape , and to suppress any acausal physics here .
source terms that have several powers of curvature in the coupling may be tuned to behave in this way , and often appear in modified gravity theories .
this approach could also give a physical picture for information to escape into a tachyonic hidden sector .
+ an alternative realization of nvnl would only allow communication between tachyonic and regular matter sectors where curvature was sufficiently large , or energy near the plank scale .
tachyons produced in high curvature areas would decay into particles of smaller negative mass and dissipate strongly in regular space , interacting only gravitationally , perhaps as dark energy .
the tachyonic hidden sector could also be seen as a realization of a baby universe model since many locally positive energy modes in the baby universe may appear tachyonic when extended back into ours .
space - time in these scenarios is generally nonsingular .
they may or may not have apparent horizons or long lived but temporary remnants . some share the same problem with other horizon structure models that horizon formation is nonlocal .
the main concern is then the deviation from classical gravity at low curvature , as well as perhaps significant changes to stellar physics .
the messy business of finding string solutions that look like fuzzballs is a bit beyond the scope of this paper , but interesting questions can still be asked .
could stable stars form around fuzzballs or fire - bubbles ? some models with nonsingular spacetimes involves long lived remnants and the back reaction of a negative energy flux into the hole@xcite .
singularity theorems are evaded by violating the energy conditions , with newer models replacing the long lived remnant with a bouncing black hole that explodes before the page time . to understand this model
, we can look at it as the natural extension of the hidden sector thermofield plus post - selection model . consider a scenario with two asymptotic regions connected by a posts - selected singularity .
all of the information of the infalling matter , as well as the entangled partners of hawking pairs , can be ordered to end up in the second region .
we do nt want any losses from evaporation into the second region , and we do want to transfer the entanglement to outgoing hr .
er = epr attempts to fix these issues by identifying the ` dumping ground ' surface with the states of outgoing hr , by using wormholes .
essentially it is cutting out the middle man .
the second asymptotic region is replaced with a complex ` hydra ' space - time that diverts the information to the mouths that are outgoing hawking pairs .
the final state boundary condition inside the hydra allows information to escape from the body(black hole ) to the heads(hr ) .
this is functionally equivalent to ordinary post - selection scenarios , with the added difficulty of the nontrivial interior hydra space - time .
it is hard to see what other predictions might be extracted from this proposal .
the scenario of entanglement emergent spacetime should be seen as a distinct model from evaporation by post - selected wormhole pairs .
the exploration of emergent spacetime models deserves a paper of it s own@xcite .
complementarity is not a theory so much as a characteristic common to many models .
these models attempt to have the best of both worlds , and are the main target of amps and similar arguments . here
we have observer dependent structure at the horizon . in one frame matter
is deflected away by a stretched horizon .
the amps argument and nice - slice argument are difficult problems since they consider frames that bridge the two complementary descriptions , attacking the basic premise that only individual observers descriptions need to be consistent .
if observers that could detect violations of unitarity can only do so in a transplankian regime , this is thought to be good enough .
it s added to models with horizon structure to hide that structure from local observers .
fuzzball complementarity claims you do not notice hitting a fuzzball at low curvature , due to it having so many degrees of freedom .
one could say the same thing about a baby universe theory , and call it baby complementarity , though this would be unnecessary .
one comment made by hawking shortly after his concession was that unitarity would be preserved by considering superpositions of spacetimes .
conversely he also said that information would still be lost when one is restricted to a particular classical black hole background .
most discussion of firewalls has taken place on such a classical background .
the claim here is that the paradox can be circumvented by distributing the entanglement between early and late hr over a large number of geometries . by doing
so it becomes exponentially harder for an observer outside the hole to reconstruct the copy of the emitted state needed to observe a violation of entanglement monogamy by crossing the horizon .
alternatively , we can have the hawking pairs deviate from the normal statistics by a similar exponentially small amount , making the firewall that would result from such vacuum scrambling very weak .
the small deviations and the eventual restoration of unitarity are left to quantum gravity .
the primary assumption seems to be that all infalling states must be decohered with respect to a basis that spans the different background geometries , and that an observer who measures the entire state including such geometric correlations can not fall through a particular horizon@xcite . without quantum gravity to treat the superposition of classical geometries , it is difficult to analyze such a scenario .
one observation here is that it predicts the apparent loss of correlation for the late time radiation by decohering it among large number of classical geometries .
this is most of the reason why the information paradox has seen a revival .
an elegant thought experiment by amps , along the same lines as old no - cloning objections to complementarity led to the postulation of a new entity that would thwart any attempt to detect cloning inside the horizon .
that would be called a firewall .
different models attempt to either avoid or generate firewalls in different ways .
blue shifted infalling matter from a second asymptotic region , does so in some models that have such regions . in string models such as fuzzballs ,
it has been merged with the idea of the stretched horizon . in some others
, it may be a consequence of the junction between interior and exterior fields . hawking radiation consists of vacuum pairs that should be maximally entangled upon creation .
the outgoing part of that radiation should eventually be part of a pure state , so it appears that entanglement entropy of the black hole must at some point decrease to zero .
this requires entanglement between radiation at different times or other matter entangled with the collapsing matter . according to amps ,
this violates the monogamy of maximal entanglement . to prevent this ,
everything is scrambled by the firewall before it can be observed .
+ the principle of monogamy of entanglement results from the principle of no - cloning . firewall and cloning
paradoxes are essentially the same . consider a black hole built out of one half of a large ensemble of epr pairs . upon collapse entanglement entropy
is already maximal , and should begin decreasing immediately with the emission of hawking radiation .
the mechanism whereby the horizon knows when to form a firewall or emit quanta correlated with some newly arriving matter is still unknown .
there seems to be no reason they should appear at any particular distance from the horizon , save that we do nt want to be able to detect them .
the formation of a radially static structure inside the horizon makes even less sense , as that would be a space like surface .
the problem of apparent vs. global horizons makes it clear that they form from nonlocal effects .
lastly in models with an interior spacetime , an energetic firewall would appear to violate the field equations of gr .
a non - exhaustive table of popular models .
tos = trace over states ; tfd = thermofield double ; bhfs = black hole final state ; bu = baby universe ; ths = tachyonic hidden sector ; nvnl = nonviolent nonlocality ; tfps = thermofield double plus postselection ; epr = wormhole pair production model ; fb = fuzzballs ; msd = macroscopic entanglement decoherence ; rem = remnants + there are even more variations between these scenarios .
one could imagine remnants that only partially restore purity but have shorter lifetimes .
nvnl is a large class of potential models some of which might restore approximate locality and be unitary .
* unitarity is usually believed to follow from the ads / cft duality applied on a pre - collapse surface , combined with reversing the ordinary unitary collapse evolution .
this is put to use explicitly in the pull back / push - forward method . *
* there are versions with and without firewalls for tfd and epr .
one can also use a version of postselection to exclude firewalls specifically@xcite .
+ it seems that black hole physics is again in a golden age for theorists , but maybe just an age of headaches for everyone else .
ahmed almheiri , donald marolf , joseph polchinski , james sully , _
`` black holes : complementarity or firewalls?''_. ( 2012 ) arxiv:1207.3123 [ hep - th ] ; cf .
s. l. braunstein , `` black hole entropy as entropy of entanglement , or it s curtains for the equivalence principle , '' arxiv:0907.1190v1 [ quant - ph ]
|
the past year has seen an explosion of new and old ideas about black hole physics . prior to the firewall paper ,
the dominant picture was the thermofield model apparently implied by ads / cft duality@xcite .
while some seek a narrow response to almheiri , marolf , polchinski , and sully,(amps)@xcite , there are a number of competing models .
one problem in the field is the ambiguity of the competing proposals .
some are equivalent while others incompatible .
this paper will attempt to define and classify a few models representative of the current discussions .
|
nearly all digital signal - measuring equipment found in laboratories throughout the world , from the humble voltmeter up to powerful computerized data - logging systems and high - frequency digitizing scopes , operate with an internal master clock that sets the pace for an analog - to - digital converter that translates the electrical output of a transducer into a bit pattern .
the master clock frequency is usually held as stable as possible and the sample intervals are fixed to a very high degree of precision .
correspondingly , most signal - analysis techniques are meant to be used on evenly spaced data : this is true for the discrete fourier transform ( dft ) and also for autoregressive ( ar ) or moving average ( ma ) modeling of data @xcite .
however some data happen to be unevenly sampled : this is especially true for astronomers , who are seldom so lucky as to have an uninterrupted series of clear nights , and in general are bound to observe whatever comes from the sky , whenever it comes , and have to search for periodicities amid these scattered data .
indeed it was the astronomical community that developed the first effective spectral estimation techniques for unevenly sampled data @xcite .
uneven sampling has special properties : beutler proved rigorously @xcite that in general uneven sampling is not band - limited and later showed that a random , poisson - distributed set of sampling times allows perfect signal reconstruction @xcite .
earlier , yen @xcite was able to derive modified forms of the shannon reconstruction formula for different types of uneven sampling , which are however much more complex than the corresponding formula for even sampling . and
indeed , randomly sampled signals are not easy to analyze and
many standard methods must be abandoned , although in some cases one can restore regular sampling using reconstruction algorithms @xcite . since no sampling clock is quite perfect and is normally affected by noise @xcite and by deterministic drifts ( that may be periodic ) , all regular sampling should actually be regarded as quasi - regular sampling .
ignoring this may be dangerous , because it is clear that quasi - regular sampling of a deterministic signal ( e.g. a sinusoidal signal ) may introduce unwanted harmonics in the dft analysis of the sampled signal , unless corrective measures are taken .
but what happens if one samples pure noise ?
can one still produce artifacts ? if this were the case then an unrecognized quasi - regular sampling of a noisy background might become a problem since it would produce fake signals that could be mistaken for true . in this paper
i show that this is just what happens in some cases of colored noise , i.e. , in the case of the long - memory noise processes @xmath0 with @xmath1 , while the correlation between samples for noises with @xmath2 is insufficient to produce replicas of the low - frequency peak of the noise spectrum . a proof is given in section [ proof1 ] , while section [ sim ] illustrates numerical results that confirm the theoretical analysis , and section [ disc ] discusses some implications of these findings .
it is well known that a signal sampled @xmath3 times in the time interval @xmath4 can be modeled by a sum of @xmath3 exponentials and that this is equivalent to a dft : in other words the dft is a physical model of the signal @xcite and we can write : @xmath5 where @xmath6 denotes the @xmath7-th sample , and the fit coefficients @xmath8 correspond to the dft and can be calculated from the formula @xmath9 with the usual regular sampling intervals @xmath10 , the @xmath7-th sampling time is @xmath11 , the total sampling time is @xmath12 , and equation ( [ dft ] ) can be rewritten as follows : @xmath13 if sampling is not quite regular , the sampling times @xmath14 are replaced by @xmath15 , and equation ( [ dft2 ] ) becomes @xmath16 now we assume a periodic sampling pattern with a period equal to @xmath17 clock ticks , so that @xmath18 and @xmath19 ; we also assume at first that @xmath20 is an integer ( numerical simulations show that this requirement can be relaxed and @xmath20 can be real ) and we expand the relative timing shift @xmath21 as a fourier sum : @xmath22 and in addition , we assume the relative timing shifts to be very small , i.e. , @xmath23 .
next we notice that @xmath24 , and then , using equations ( [ dft2uneven ] ) and ( [ rts ] ) and after a few cumbersome but straightforward passages , we can approximate the observed dft with the following formula : @xmath25 moreover if we make the rather weak assumption that the phase of the noise dft in different frequency bins is uncorrelated so that @xmath26 if @xmath27 where @xmath28 is the usual ensemble average , then the dft ( [ obsdft ] ) gives the following spectrum : @xmath29 ( the hypothesis of phase independence is quite common , because it is essential for noise generators like that of timmer and knig @xcite , and is supported by the numerical results reported in @xcite ) . if the noise is white , i.e. , the spectral density is flat , or if it is a @xmath0 noise with a spectral index @xmath30 , we see from eq .
( [ obspsd ] ) that the periodic uneven sampling amounts to the addition of a ( small ) non - flat background .
in fact a @xmath0 noise has a discrete spectrum @xmath31 , therefore the observed spectrum ( [ obspsd ] ) becomes @xmath32 however , if the spectral index @xmath33 is greater than 2 , then the @xmath34-th harmonic of the relative time shift @xmath21 produces a peak over the power - law background , which is just the low - frequency noise peak , shifted to the @xmath35-th frequency bin .
the analysis that leads to eq .
( [ obspsd ] ) assumes small relative timing shifts , but in this section i report numerical simulations carried out with the exact power - law noise generator described in @xcite that do support the analytical results also for large relative timing shifts @xcite .
the generator used in the simulation runs produces power - law noise from a superposition of random exponential pulses , and is exact in the sense that it produces a process that is theoretically guaranteed to yield a range - limited power - law spectrum between two extreme ( angular ) frequencies @xmath36 and @xmath37 .
the generator takes correctly into account the correlation between samples in colored noises , and works also with unevenly spaced sampling times . in these simulations , time is in arbitrary units , and the average sampling interval is @xmath38 arb .
units ; the choice of time units also sets the corresponding frequency units used for the relaxation rates @xmath39 and @xmath37 .
figure [ fig1 ] shows a simulated signal obtained with the noise generator for a @xmath40 noise ; in this case the generator parameters are @xmath41 , @xmath42 , and @xmath43 , i.e. , the spectrum has a power - law region @xmath40 that spans the frequency interval @xmath44 , and the pulse rate has been set at @xmath45 pulses per unit time , so that the resulting noise signal is gaussian to a very high degree @xcite . in this case the sampling time has been sinusoidally modulated : @xmath46 ( the period for uneven sampling is 4 samples ) , and there are in all @xmath47 samples . figure [ fig2 ] shows the dft spectrum of the signal of figure ( [ fig1 ] ) : a comparison with the exact theoretical spectrum of the noise generator @xcite @xmath48\end{aligned}\ ] ] which has a @xmath0 power - law in the range @xmath49 shows that on the whole the sampled noise process produced by the noise generator actually behaves as predicted by theory @xcite , except for a small peak at the frequency of the sampling time modulation .
this small peak only shows up in this and in other runs ( not shown here , but easily reproducible @xcite ) with @xmath1 , and this lends support to the proofs of the previous sections .
the spectra shown in this and in the other figures have been partly detrended with a hanning window ( a general introduction to the need of the detrending step can be found , e.g. , in @xcite ; see also the qualitative considerations in @xcite ) . a closer look at the modulation peaks yields however a much more striking confirmation of the analytical results : in fact the theoretical spectral density ( [ psdplaw ] ) of the noise generator has a @xmath0 power - law region for @xmath49 , while for @xmath50 it has a @xmath51 behavior , and this means that from equation ( [ obspsd ] ) we expect that the correction term due to uneven sampling is negligible just at the modulation frequency , while there should be two side - peaks whose exact shape depends on the low - frequency limit of the @xmath0 region , i.e. , on @xmath36 . and
indeed this is just what happens in the simulations , as shown in figure [ fig3 ] , where part * a. * shows the region of the averaged spectrum in figure [ fig2]b close to the peak due to sampling time modulation .
figure [ fig3]b shows the expected behavior calculated from equation ( [ obspsd ] ) and from the conditions used in the generation of the noise process and listed above ( the correction for the incoherent gain of the hanning window is also included ) . finally , figure [ fig3]c is the superposition of parts [ fig3]a and [ fig3]c , where we see that the calculated shape closely matches the observed shape
. figure [ fig4 ] shows the results of a similar calculation performed on the spectrum of a single realization of the noise process shown in figure [ fig2]a : figure [ fig4]a is the zoomed portion of the spectrum around the modulation peak .
figure [ fig4]b is a still closer zoom , and the arrow in the figure shows the position of the modulation frequency : notice that there is no peak just where one would naively expect to find one . the low frequency part of the spectrum in figure [ fig2]a has been inserted in equation ( [ obspsd ] ) to obtain the spectrum in figure [ fig4]c , and we see that there is an almost perfect correspondence with the peak in figure [ fig4]c : this means that the correlation terms between different frequencies ( the cross - terms in the derivation of equation ( [ obspsd ] ) ) are negligible even for a single realization of the noise process .
i have also noted that there must be a dependence of the split - peak shape on the exact shape of the low - frequency part of the spectrum : figure [ fig5]a shows the shape of the peak for a larger value of @xmath36 ( @xmath52 ) . in this case
the low - frequency part of the noise spectrum ( [ psdplaw ] ) has a wider @xmath51 region , and the side - peaks must be correspondingly lower and further apart : indeed this is just what happens in figure [ fig5]a . a further confirmation is provided by figure [ fig5]b , which shows the peak for a smaller value of @xmath36 ( @xmath53 ) : the side - peaks are much higher and also closer .
while most of the observed power - law ( @xmath0 ) noises have spectral indexes @xmath54 , with an apparent clustering around @xmath55 , red noises , i.e. , noises with spectral indexes @xmath1 , also show up in several unrelated systems @xcite like the water level of the nile river , economics , orchid population size @xcite and local temperature fluctuations and affect precise timekeeping @xcite and our ability to predict environmental and animal population variables @xcite .
noises with @xmath1 also appear in the energy level fluctuations of quantum systems @xcite and in timing noise in pulsars @xcite .
because of their extreme peaking behavior at low frequencies these noises are also called `` black '' @xcite , and they display marked persistence properties @xcite that may lead to the mistaken identification of underlying trends in experimental data @xcite . from the results reported in this paper
it follows that these noises pose yet another potential danger to experiments that use uneven sampling , because their long - memory properties give rise to artifacts in the dft spectra
. however equation ( [ obspsd ] ) also shows that the single peak in the case of true @xmath0 noise or the side - peaks in the case of range - limited @xmath0 noise can be modulated both in amplitude and in frequency by modulating either the noise spectrum or the relative timing shift amplitude , or the repetition index @xmath20 : this means that uneven sampling of colored noise could be utilized to encode information , and since an encoding noise appears at first sight just ordinary noise , this could be used to implement a secure communication channel ( there is a very rich literature on this topic , but here i give only a reference to a classic book @xcite and to a recent paper @xcite ) .
a simple example of the kind of modulation that can be achieved can be gleaned from figures [ fig3 ] , [ fig4 ] and [ fig5 ] : if one uses the noise generator @xcite , it is possible to modulate the shape of the low - frequency part of the spectrum with a proper change of @xmath36 and in this way one modulates in turn both the amplitude and the position , i.e. , the frequency , of the side peaks .
all the simulations reported here have been produced with a slightly modified version of the example program in the _ plnoise _ noise generator described in @xcite and @xcite and available from the cpc program library : http://www.cpc.cs.qub.ac.uk / cpc/. the main change in that code amounts to the substitution of the instruction that advances time at each sampling step this figure show a noise signal produced with the generator described in @xcite .
the parameters in this specific run are @xmath41 , @xmath42 , and @xmath43 , i.e. , the spectrum has a power - law region @xmath40 that spans the angular frequency interval @xmath56 .
the generator produces power - law noise from a superposition of random exponential pulses , and in this run the pulse rate has been set at @xmath45 pulses per unit time , so that the resulting noise signal is gaussian to a very high degree @xcite .
time is in arbitrary units , and the average sampling interval is @xmath38 ( arb .
units ) ; the choice of time units also sets the corresponding frequency units used for the relaxation rates @xmath39 and @xmath37 .
the sampling time has been sinusoidally modulated : @xmath46 , and the signal has been sampled @xmath47 times .
part * a*. shows the whole signal generated in this run ( time does not start from zero , because at the beginning some samples are used for the generator initialization and are discarded ) ; part * b*. shows the initial 32768 valid samples .
notice that even though the sampling time modulation is rather large ( 20% ) , it is quite invisible in the zoomed figure . , width=480 ] * a. * spectrum of the signal shown in figure ( [ fig1 ] ) vs. the scaled angular frequency @xmath57 .
the arrow marks the peak from sampling time modulation at the expected position @xmath58 , where @xmath59 is the ( angular ) nyquist frequency and @xmath60 is the sampling frequency .
* b. * spectrum averaged over 16 realizations of the same noise process : the solid line shows the expected ( theoretical ) behavior of the noise spectrum @xcite , corrected for the incoherent gain of the hanning window that has been used for trend removal .
the upward bend at high frequency in the spectrum * b * is due to aliasing which becomes apparent after averaging , while the upward bend at low frequency is due to the uncorrected dc component which can not be removed by windowing .
, width=480 ] * a*. ) averaged spectrum of figure [ fig2]b zoomed and centered on the position of the small peak ( i.e. @xmath61 ) , with linear scales on both axes . *
b*. ) expected behavior calculated from equation ( [ obspsd ] ) and from the conditions used in the generation of the signal in figure [ fig1 ] ( the correction for the incoherent gain of the hanning window is also included ) .
* c*. ) superposition of parts * a * and * b * : we see that the calculated shape closely matches the observed shape * b*. , width=384 ] * a*. ) spectrum of figure [ fig2]a zoomed and centered on the position of the small peak ( i.e. @xmath61 ) , with linear scales on both axes .
* b*. ) an even closer view of the peak shows that it is actually a split peak : the arrow marks the position of the modulation frequency . *
c*. ) this part shows what one obtains if one uses equation ( [ obspsd ] ) and the low - frequency part of the spectrum of figure [ fig2]a to calculate the expected shape of the peak .
the calculation includes the correction for the incoherent gain of the hanning window that has been used for trend removal .
we see that the calculated shape * c*. matches very well the observed shape * b*. , width=384 ] this figure illustrates how the peak splitting changes for different noise shapes : * a*. ) in this case the noise generation parameters are the same as for the signal in figure [ fig1 ] , except for the lowest relaxation rate @xmath62 , and the figure shows the averaged spectrum zoomed and centered on the position of the small peak .
the softer low - frequency behavior in noise spectrum produces side peaks that are smaller and further apart than those shown in figure [ fig3 ] .
the solid line shows the expected behavior , calculated as in figure [ fig3]b * b*. ) a smaller value of the lowest relaxation rate , @xmath63 yields instead much closer and higher side peaks .
once again we see that the calculated shape ( solid line ) matches very well the observed shape . because of the much narrower splitting , the scale in part * b *
is expanded with respect to * a * to improve visibility .
, width=384 ]
|
the vast majority of sampling systems operate in a standard way : at each tick of a fixed - frequency master clock a digitizer reads out a voltage that corresponds to the value of some physical quantity and translates it into a bit pattern that is either transmitted , stored , or processed right away . thus signal sampling at evenly spaced time intervals is the rule : however this is not always the case , and uneven sampling is sometimes unavoidable .
while periodic or quasi - periodic uneven sampling of a deterministic signal can reasonably be expected to produce artifacts , it is much less obvious that the same happens with noise : here i show that this is indeed the case only for long - memory noise processes , i.e. , power - law noises @xmath0 with @xmath1 .
the resulting artifacts are usually a nuisance although they can be eliminated with a proper processing of the signal samples , but they could also be turned to advantage and used to encode information .
|
the observational evidence inferred from the spatial variations of the cosmic microwave background ( cmb ) temperature suggests that the curvature perturbations that are present prior to decoupling ( but after equality ) must be predominantly adiabatic and gaussian with an ( almost ) scale - invariant spectrum @xcite .
a mode is defined to be ( predominantly ) adiabatic if the fluctuations of the entropy density are subleading after matter - radiation equality ( but prior to decoupling ) .
an exactly adiabatic mode of curvature fluctuations may be produced , for instance , by single - field inflationary models .
if this is the case , the scalar fluctuations of the inflaton field ( with wavelength larger than the hubble radius ) are transferred to curvature fluctuations whose value stays ( almost ) frozen from the early stages of the evolution of the universe until the pre - decoupling epoch when the minute spatial variations of the geometry are imprinted , via the sachs - wolfe effect , on large - scale temperature inhomogeneities .
in addition to adiabatic fluctuations there may be one or more non - adiabatic modes sometimes named ( with a somehow confusing terminology ) isocurvature modes . the specific number of ( regular )
isocurvature modes ( up to four ) is determined by the physical properties of the plasma where curvature fluctuations evolve . a mode , in the present terminology
, simply means a consistent solution of the governing equations of the metric and plasma fluctuations , i.e. a consistent solution of the perturbed einstein equations and of the lower multipoles of the boltzmann hierarchy .
consider as an example the physical system formed by two fluids , i.e. radiation and pressureless matter .
this is a pretty good approximation in the description of pre - decoupling physics if the radiation is identified with the sum of photons and neutrinos and the pressureless fluid accounts for the cold dark matter ( cdm ) component . if this is the case , the plasma suports not only an adiabatic mode , but also a non - adiabatic mode for which the entropy fluctuations of the cdm - radiation system do not vanish @xcite .
it is therefore possible that the pre - decoupling curvature fluctuations have a dominant ( adiabatic ) contribution and a subdominant ( entropic ) correction .
sticking to the simplified situation described above , the large - scale contribution to the angular power spectrum would be determined by three pre - decoupling power spectra , i.e. the spectrum of the adiabatic mode , the spectrum of the entropic mode and the spectrum of their cross - correlation .
this possibility leads to a rich parameter space and to an interesting phenomenological framework ( see , for instance , @xcite ) .
as an example in ref .
@xcite a recent likelihood analysis determined the isocurvature fraction allowed by wmap data ( as well as by large scale structure data ) by taking into account both the adiabatic and the entropic contribution of the cdm - radiation system .
the study of isocurvature and entropy modes is often conducted within a two - fluid system because of the inherent mathematical complications related with the possible proliferation of non - adiabatic modes .
it will be demonstrated here that the presence of a large - scale magnetic fields during the pre - decoupling epoch affects the whole system of curvature and entropy perturbations both qualitatively and quantitatively .
if the origin of large - scale magnetic fields is primordial ( as opposed to astrophysical ) it is plausible to expect that the pre - decoupling plasma is effectively magnetized .
different studies have been devoted , in recent years , to constrain a hypothetical primeval magnetic field from cmb studies ( see @xcite for a short review and @xcite for a more extended discussion ) . indeed
, large - scale magnetic fields that are present in the pre - decoupling epoch may affect the polarization power spectra ( both the ee , bb and eb angular power spectra ) either directly @xcite or through the faraday effect @xcite that may rotate an initial e mode ( produced thanks to the polarized nature of thompson scattering ) into a b mode . barrow and subramanian , in a series of papers @xcite , analyzed the vector modes ( and the polarization power spectra ) induced by large - scale magnetic fields .
recently @xcite a numerical approach has been developed to analyze vector and tensor modes .
vector , as well as tensor modes have been discussed also in @xcite .
while the effect of vector modes will be crucial at small scales ( i.e. @xmath0 ) the contribution of tensor modes may be relevant also for larger scales ( i.e. @xmath1 ) .
large - scale magnetic fields induce vector and tensor modes but also various scalar modes whose classification and analysis will be one of the themes of the present investigation . more specifically , large - scale magnetic fields affect the usual adiabatic and entropic modes @xcite that may be present in the pre - decoupling epoch .
in fact , the divergence of the magnetohydrodynamical lorentz force influences the tight - coupling evolution of photons and baryons .
moreover , the magnetic fields gravitate .
therefore , they contribute to the hamiltonian constraint and ( in a weaker fashion ) to the momentum constraint . in @xcite , an interesting covariant approach for the analysis of cosmological perturbations induced by large - scale magnetic fields has been proposed and further developed in @xcite .
the approach used in @xcite was to assume a background magnetic field which uniform but , at the same time , sufficiently weak not to break the isotropy of the background geometry .
the present approach is more similar to the one of ref .
@xcite where the magnetic field has been taken fully inhomogeneous .
moreover , the problems related to the present analysis concern the specific modifications induced by large - scale magnetic fields on the adiabatic and non - adiabatic modes that are normally used to set initial conditions for cmb anisotropies .
this goal requires a full description of the magnetized photon - baryon system in the presence of neutrinos and cdm particles both before and after decoupling .
the strategy would then be to study how large - scale magnetic fields affect the system of equations describing adiabatic and non - adiabatic modes before the time of radiation - matter equality , i.e. deep in the radiation - dominated epoch .
the magnetized solutions will be then evolved both analytically and numerically across the radiation - matter transition and up to the time of photon decoupling .
as mentioned , large - scale magnetic fields are assumed here to be fully inhomogeneous and characterized by an appropriate two - point function . in this way
the spatial isotropy of the background geometryis not broken and the line element can then be written , in the spatially flat case , as @xmath2 , \label{metric}\ ] ] where @xmath3 is the conformal time coordinate and @xmath4 is the scale factor . defining the appropriately rescaled magnetic fields as @xmath5 ( see section2 ) , the relevant two - point function can be written ( see the appendix for a summary on the conventions ) as @xmath6 where @xmath7 is the magnetic field power spectrum and @xmath8 .
large - scale magnetic fields will be treated here in a magnetohydrodynamical ( mhd ) approximation @xcite in curved space - time .
this description is justified by different considerations .
in fact , before decoupling , not only the plasma is globally neutral but the value of the debye scale much smaller than the hubble radius and it is also much smaller than the electron and photon mean free path @xcite . as far as the magnetic properties of the system are concerned
, the plasma behaves , in this situation , as a single ( intrinsically charged but globally neutral ) fluid where the total ohmic current is solenoidal and the ohmic electric fields are suppressed by the finite value of the conductivity .
the same kind of mhd description has been used also in the analysis of the vector and tensor modes of the geometry ( see , for instance , @xcite and @xcite ) .
it should be borne in mind , however , that it would be incorrect to use the same kind of mhd approximation in the analysis of faraday effect ( as sometimes assumed ) .
faraday effect concerns the propagation of electromagnetic waves in a cold ( or warm ) plasma and can not be treated within mhd which is valid for typical frequencies much smaller than the plasma frequency and typical scales much larger than the debye scale .
so in full analogy with the discussion of faraday effect in the laboratory ( see for instance @xcite ) a two - fluid description is mandatory @xcite .
a direct consequence of the mhd approach is that magnetic flux and magnetic helicity will be , to a good approximation , conserved in the limit of large conductivity @xcite . in the friedmann - robertson - walker background ( [ metric ] )
, flux conservation implies that the magnetic field strength increases as @xmath9 as we go back in time . at the present time magnetic fields
are observed , at the @xmath10 g level , in galaxies @xcite , clusters @xcite and in some superclusters @xcite .
even if the fields have different properties in diverse classes of structures and even if the determination of magnetic fields in superclusters is still in progress , it is rather plausible to admit that we live in a sort of magnetized universe @xcite . in the case of galaxies , if the magnetic field present prior to gravitational collapse was of the order of @xmath11 ng , compressional amplification alone could have produced the observed magnetic field even without the need of a dynamo action . in recent years
interesting progresses have been made in the analysis of the dynamo problem especially in connection with the role of global conservation laws of magnetized plasmas @xcite .
therefore , the protogalactic field could be much smaller than @xmath11 ng and still be explain the observed magnetic fields in galaxies , while in this framework cluster magnetic fields are related to the way dynamo saturates .
aside from this consideration , if we blue - shift a protogalactic magnetic field ( coherent over a typical mpc scale ) of @xmath11 n g at the epoch of photon decoupling we will end up with a magnetic field of mg strength that can potentially affect cmb anisotropies . the mechanisms that may produce magnetic fields will not be specifically examined and we refer the reader to a recent revies @xcite . however , most of the mechanisms predict a stochastically distributed magnetic field that does not break spatial isotropy
. it will then be interesting to scrutinize the implications of a pre - decoupling field on the scalar cmb anisotropies . in this sense
the ambition of the present approach is a model - indepedent approach that is similar to recent attempts of constraining mixtures of adiabatic and non - adiabatic modes in spite of their specific origin ( see @xcite and references therein ) . to be more specific , the strategy will be to treat the magnetic fields as a further element of the primordial plasma that may contain both adiabatic ( or entropic ) and non - adiabatic modes . in doing a gauge invariant description
will be adopted since , as it will be illustrated , a simpler description of the matter - radiation transition can be achieved .
to say anything concerning cmb anisotropies we have to estimate what is the value of curvature perturbations after equality ( but before decoupling ) and in the long wavelength limit . , the long wavelengths , in the present terminology , are the ones satisfying @xmath12 . ]
such an estimate , is , unfortunately , not available in the present literature .
it is then appropriate to introduce the description developed in this paper by defining the density contrast on uniform curvature hypersurfaces , conventionally denoted by @xmath13 @xcite @xmath14 where the prime denotes a derivation with respect to @xmath3 and @xmath15 ; @xmath16 is the ( total ) background fluid density while @xmath17 is the pressure .
similarly , in eq .
( [ zetadef ] ) , @xmath18 , @xmath19 is the fluid density fluctuation in the uniform curvature gauge @xcite and @xmath20 is the fully inhomogeneous magnetic energy density which is gauge - invariant and quadratic in the magnetic field intensities . if the fluid has different consituents ( for instance , in one of the simplest cases , dusty matter and radiation ) , then the total density contrast on uniform curvature hypersurfaces , i.e. @xmath21 can be written as the weighted sum of the individual fluid perturbations and of the magnetic field density contrast , i.e. @xmath22 given the form of eqs .
( [ zetaidef1 ] ) the relative fluctuations in the entropy density @xmath23 can be defined as @xmath24 in this language , the adiabatic mode corresponds to the situation where @xmath25 and @xmath26 for each fluid of the mixture .
the other important variable treated here will be gauge - invariant curvature fluctuation .
clearly , in different gauges , the physical meaning of @xmath27 may be different . ]
usually denoted by @xmath27 , and related to @xmath13 as @xmath28 where @xmath29 is the ( gauge - invariant ) bardeen potential . in the long wavelength limit , the second term at right hand side of eq .
( [ zetator ] ) can be neglected and @xmath13 is approximately equal to @xmath27 .
when the plasma is constituted by a mixture of relativistic fluids , the evolution of @xmath13 can be written , in the long wavelength limit , as @xmath30 where @xmath31 is the non - adiabatic pressure density variation accounting for the contribution of entropy fluctuations ( see eq .
[ dpnad ] ) . equation ( [ zetacons ] ) contains , at the right hand side , not only the contribution of @xmath31 but also the effect of the inhomogeneous magnetic field .
the logic followed in the present investigation will then be to provide a fully gauge - invariant system of equations where the evolution of magnetized curvature and entropy modes can be characterized as regular solutions deep in the radiation - dominated regime .
then the various modes will be followed across the radiation - transition both analytically and numerically . in section 2
a gauge - invariant system of governing equations will be proposed for the simultaneous description of magnetized adiabatic and entropy modes . in section 3
the analysis of various magnetized modes will be presented in the realistic situation where the plasma is constituted by a mixture of baryons , neutrinos , cdm and photons.the fate of the regular magnetized modes across the radiation - matter equality will be discussed in section 4 .
the precise amplitude of the curvature perturbations before decoupling will then be computed for the wavelengths that are larger than the hubble radius after equality . in section 5
some more phenomenological considerations on the properties of the sachs - wolfe plateau will be discussed with the aim of providing a possible starting point for more specific numerical discussions .
finally , section 6 contains the concluding discussion and a summary of the possible developments of the present investigation .
a series of technical results have been collected in the appendix in connection with the proper parametrization of the various power spectra arising in the problem .
the system of magnetized curvature perturbations can be obtained by considering the case of a gravitating charged fluid which is , however , globally neutral for scales larger than the debye length . around the time of decoupling , for a typical temperature of @xmath32 ev and for an electron density @xmath33 the debye length is of the order of @xmath34 , i.e. much smaller than the hubble radius at the corresponding epoch .
the background equations ( and the background metric ( [ metric ] ) ) are not affected by the presence of magnetic fields and have the standard form in terms of the ( total ) energy and pressure densities , i.e. @xmath16 and @xmath17 : @xmath35 denoting with @xmath36 the first - order ( scalar ) fluctuation of the corresponding quantity , the perturbed einstein equations can be written as @xmath37 where @xmath38 is the einstein tensor while @xmath39 and @xmath40 are , respectively , the energy - momentum tensors of the fluid perturbations and of the electromagnetic sources . the perturbed covariant conservation equations and by the equations for the maxwell field strength can instead be written as @xmath41 in eqs .
( [ p1])([p3 ] ) , @xmath42 denotes the covariant derivative defined from the metric @xmath43 defined in eq .
( [ metric ] ) . in eq .
( [ p3 ] ) @xmath44 is the ohmic current .
the ( total ) velocity field , @xmath45 , obeys , as usual @xmath46 .
recalling that @xmath47 and @xmath48 the evolution equations for the rescaled electric and magnetic fields become : @xmath49 where @xmath50 the one - fluid mhd description adopted here @xcite amounts then to the requirement that both the maxwell fields and the induced ohmic current are solenoidal @xmath51 which implies that @xmath52 since we will be interested in effects occurring over length - scales much larger than the debye scale , the displacement current can be safely neglected in eq .
( [ e3 ] ) .
note that @xmath53 appearing in eq .
( [ el1 ] ) is the bulk velocity of the plasma . in mhd ,
the bulk velocity of the plasma is essentially the center of mass velocity of the baryon - electron system .
therefore , @xmath53 coincides , in practice with the baryon velocity field . in eq .
( [ el1 ] ) , @xmath54 is the conductivity . in a relativistic plasma @xmath55 while in a non - relativistic plasma @xmath56 where @xmath57 is the electron mass and @xmath58 is common temperature of electrons and baryons .
the evolution equations for the coupled system defined by eqs .
( [ p1 ] ) , ( [ p2 ] ) and ( [ p3 ] ) can then be derived . for the present purposes it is useful to employ the uniform curvature gauge @xcite where the perturbed elements of the metric tensor are @xmath59 the covariant conservation of the total energy - momentum tensor , i.e. eq .
( [ p3 ] ) implies then two separate equations one for the total velocity field and the other for the density fluctuation .
these two equations can be written as @xmath60 where @xmath61 is the divergence of the ( total ) velocity field and where @xmath62 is the ( total ) sound speed .
to derive eqs .
( [ con1 ] ) and ( [ con2 ] ) from eq .
( [ p2 ] ) , the perturbed components of the energy - momentum tensor of the fluid have been parametrized as @xmath63 in eq . ( [ velp ] )
the term @xmath64 represents the contribution of the anisotropic stress .
the energy - momentum tensor of the electromagnetic sources is , instead , @xmath65 equation ( [ p1 ] ) implies two constraints ( stemming from the ( @xmath66 ) and ( @xmath67 ) components ) .
the hamiltonian and momentum constraints are then , in the off - diagonal gauge , @xmath68 = - 4\pi g a^2 [ ( p_{\rm t } + \rho_{\rm t } ) \theta + \delta { \cal v}_{\rm b } ] , \label{mom0}\end{aligned}\ ] ] where @xmath69 are , respectively , the magnetic energy density and the poynting vector . notice that @xmath70 , in the one - fluid mhd approximation , contains two terms : the first is suppressed by the conductivity , the second is of higher - order . in fact , using the second relation of eq .
( [ el1 ] ) , @xmath71 } { 4\pi a^4}. \label{int1}\ ] ] the @xmath72 component of eq .
( [ p1 ] ) can be written as @xmath73\delta_{i}^{j } - \frac{1}{2 } \partial_{i}\partial^{j } [ { \cal a } ' + 2 { \cal h } { \cal a } + \phi ] \nonumber\\ & & = 4\pi g a^2 [ ( \delta_{\rm s } p + \delta p_{\rm b } ) \delta_{i}^{j } - \pi_{i}^{j } - { \tilde{\pi}}_{i}^{j } ] , \label{ijeq1}\end{aligned}\ ] ] where @xmath74 . \label{ansb}\ ] ] the electromagnetic contribution to the hamiltonian constraint ( [ ham0 ] ) contains also a purely electric term .
however , by virtue of eq .
( [ el1 ] ) the electric contribution is suppressed with the respect to the magnetic one .
furthermore , terms containing the product of a quadratic ( magnetic ) contribution and of a metric ( or velocity ) fluctuation are of higher order and will then be neglected .
the system formed by eqs .
( [ con1 ] ) , ( [ con2 ] ) and ( [ ijeq1 ] ) ( supplemented by the constraints ( [ ham0 ] ) and ( [ mom0 ] ) ) mixes the four variables @xmath75 , @xmath76 , @xmath77 and @xmath78 .
the gauge - invariant generalization of the total density contrast on uniform curvature hypersurfaces , i.e. @xmath77 is given by @xmath13 , i.e. @xmath79 where @xmath80 .
on the other hand , the gauge - invriant generlization of @xmath81 is related with the curvature fluctuations on comoving orthogonal hypersurfaces , denoted by @xmath27 : @xmath82 finally , the variable @xmath76 is related with one of the two bardeen potentials , usually denoted by @xmath29 : @xmath83 the variables @xmath13 , @xmath27 and @xmath29 are all gauge - invariant and while they have been introduced , for simplicity , in the uniform curvature gauge , they have interesting physical interpretations also in different gauges . for instance , @xmath13 can be interpreted as the curvature perturbation on constant density hypersurfaces .
this interpretation is clear by looking at the gauge - invariant form of the hamiltonian constraint . inserting eqs .
( [ def0 ] ) , ( [ def1 ] ) and ( [ def2 ] ) into eqs .
( [ ham0 ] ) and ( [ mom0 ] ) the gauge - invariant forms of the hamiltonian and momentum constraints can be obtained : @xmath84 in the gauge - invariant approach presented here , the hamiltonian constraint ( [ ham1 ] ) fixes the relation between @xmath13 and @xmath27 .
if we are interested in the evolution of the system for length - scales much larger than the hubble radius we will have that the term containing @xmath85 is sub - dominant and the constraint is satisfied if @xmath86 . the momentum constraint ( [ mom1 ] ) can be used to compute the divergence of the total velocity field @xmath78 once @xmath27 and @xmath29 are known .
using eqs .
( [ def0 ] ) , ( [ def1 ] ) and ( [ def2 ] ) into eq .
( [ con1 ] ) , and recalling eqs .
( [ fl ] ) , the following ( gauge - invariant ) evolution equation for @xmath13 can then be obtained @xmath87 - \frac{\theta}{3 } + \frac{\vec{e } \cdot \vec{j}}{3 a^4 ( p_{\rm t } + \rho_{\rm t})}. \label{zetamas}\ ] ] in eq .
( [ zetamas ] ) , the pressure fluctuation of the fluid component has been split as @xmath88 , where @xmath31 is the term proportional to the intrinsic entropy fluctuations of the fluid and @xmath89 is the total sound speed .
this decomposition is usual when dealing with mixtures of relativistic fluids even in the absence of magnetic fields .
the explicit form of @xmath31 can be evaluated once the chemical composition of the fluid is specified . by introducing the individual sound speeds of the various fluids . ] and the relative fluctuations of the specific entropy in a generic pair of fluids , i.e. @xmath90 the general form of @xmath31 can be written as @xmath91 the indices appearing in eq .
( [ dpnad ] ) ( not to be confused with covariant spatial indices ) label the different components of the fluid and , by definition , @xmath92 ( see eq .
( [ defcsi ] ) .
the gauge - invariant density contrasts of each fluid , i.e. @xmath93 , can be evaluated , for instance , in the uniform curvature gauge where they correspond to @xmath94 as already introduced in eq .
( [ zetaidef1 ] ) .
the variable @xmath13 that appears in the hamiltonian constraint ( [ ham1 ] ) can be also written in terms of the individual @xmath93 supplemented by tha magnetic contribution @xmath95 : @xmath96 in connection with eq .
( [ zetamas ] ) the following remarks are in order : * even if the intrinsic ( non - adiabatic ) contribution vanishes , the evolution of @xmath13 may change over scales larger than the hubble radius since its evolution is directly sourced by the magnetic energy density ; * this new ( entropic ) contribution vanishes when the universe is dominated by radiation but becomes relevant as soon as the speed of sound @xmath97 so that , for instance , it is relevant in the context of the radiation - matter transition ; * by virtue of the momentum constraint ( [ mom1 ] ) and of the hamiltonian constraint ( [ ham1 ] ) , the contribution containing @xmath78 at the left hand side of eq . ( [ zetamas ] ) is proportional , to leading order , to @xmath98 and to @xmath99 ( both contributions are then small on super - hubble scales ) ; * as previously remarked in deriving eq .
( [ mom1 ] ) the term @xmath100 appearing in eq .
( [ zetamas ] ) is strictly zero in the superconducting limit and negligible for finite conduvctivity .
the last statement can be easily verified since , in the mhd approximation , the relations of eq .
( [ el1 ] ) imply the following chain of equalities @xmath101 now , the first term in the second equality of eq .
( [ int2 ] ) vanishes in the limit of large conductivities or , more precisely , in the limit @xmath102 where @xmath103 is the hubble parameter .
the second term in the last equality of eq . ( [ int2 ] ) is of higher order . to complete the discussion we have now to study the relations implied by eq .
( [ ijeq1 ] ) in terms of @xmath13 , @xmath27 and @xmath29 .
using eqs .
( [ def0 ] ) , ( [ def1 ] ) and ( [ def2 ] ) into eq .
( [ ijeq1 ] ) we can extract the trace - free ( gauge - invariant ) contribution and apply the differential operator @xmath104 : @xmath105 = - \frac{16\pi g
a^2 \rho_{\rm t}}{3 ( \rho_{\rm t } + p_{\rm t } ) } [ \nabla^4 \pi + \nabla^4 \tilde{\pi } ] .
\label{tfree}\ ] ] on the other hand , the trace - full contribution stemming from eq .
( [ ijeq1 ] ) simply leads to the gauge - invariant evolution of @xmath27 , i.e. @xmath106 - \frac { c_{\rm s}^2 { \cal h } \nabla^2\psi}{4\pi g a^2 ( p_{\rm t } + \rho_{\rm t } ) } + \frac{2}{3 } \frac{{\cal h}}{p_{\rm t } + \rho_{\rm t } } ( \nabla^2 \pi + \nabla^2 \tilde{\pi } ) .
\label{rmas}\ ] ] equation ( [ zetamas ] ) has the same physical content of eq .
( [ rmas ] ) .
the reason is that for typical wavelengths larger than the hubble radius , eq .
( [ ham1 ] ) implies that @xmath86 .
this statement can be verified by showing that , for instance , eq . ( [ rmas ] )
reduces to eq .
( [ zetamas ] ) when the constraints ( [ ham1 ] ) and ( [ mom1 ] ) are taken into account .
in fact , using eq .
( [ ham1 ] ) to express @xmath27 in terms of @xmath13 and taking the first time derivative we can substitute @xmath107 in eq . ( [ rmas ] ) .
this manipulation will give rise to a bunch of terms : the first one will be essentially @xmath108 while the other contain the laplacian of @xmath29 and its derivatives
. then we can use eqs .
( [ mom1 ] ) and ( [ tfree ] ) to get rid of @xmath109 .
the resulting equation will exactly be eq .
( [ zetamas ] ) which was obtained from the covariant conservation equation .
equations ( [ zetamas ] ) and ( [ rmas ] ) suggest that the magnetic field acts , effectively , as a non - adiabatic variation of the pressure density . therefore it is useful , sometimes , to think that @xmath110 where @xmath95 now obeys @xmath111
in the absence of magnetic fields , the commonly employed strategy is to use eqs .
( [ zetamas ] ) and ( [ rmas ] ) for the estimate of the values of curvature perturbations in the long wavelength limit after the equality ( denoted , in what follows , by @xmath112 ) but before decoupling ( denoted by @xmath113 ) . the magnetic contribution to eqs .
( [ zetamas ] ) and ( [ rmas ] ) change the evolution of @xmath13 and @xmath27 , as discussed in the previous section .
moreover , the contribution to the anisotropic stress arises both from the collisionless neutrinos and from the magnetic fields . in the absence of collisionless
neutrinos the magnetic fields can only induce , during radiation , a singular mode since the magnetic anisotropic stress can not be counterbalanced by the anisotropic stress of the fluid .
the solutions of the gauge - invariant system of section 2 will now be studied in the limit @xmath114 . then the different magnetized modes ( both adiabatic and non - adiabatic )
will be followed through the matter radiation transition with the purpose of computing in detail the value of the curvature perturbations before decoupling and for wavelengths larger than the hubble radius at the corresponding epoch .
the radiation - matter transition will be modeled by the interpolating scale factor @xmath115 , \label{aint}\ ] ] which is a solution of eqs .
( [ fl ] ) when the plasma contains both pressureless matter and radiation . in eq .
( [ ai nt ] ) , @xmath116 is the value of the scale factor at equality .
the values of @xmath116 and @xmath117 are fixed by the critical fractions of matter and radiation .
in particular , and the factor @xmath118 , i.e. the indetermination in the hubble constant , will be taken @xmath119 for numerical estimates . ]
@xmath120 and @xmath121 and @xmath122 are the present values of the critical fractions of matter and radiation , @xmath123 the redshift of decoupling will be taken to be @xmath124 , implying from eq .
( [ par1 ] ) that @xmath125 ( where @xmath126 ) .
the numerical values of the matter and radiation density fractions will be taken to be @xmath127 with @xmath128 concerning the set of parameters adopted in eqs .
( [ ommr ] ) and ( [ parameters ] ) few comments are in order .
the recent wmap three year data @xcite , when combined with gold " sample of snia @xcite give values compatible with the ones adopted here .
however , if wmap data alone are used we would have , instead , a slightly smaller value for the critical fraction of matter , i.e. @xmath129 . the values of the parameters chosen here are just illustrative in order to model , in a realistic fashion , the radiation - matter transition .
therefore , any difference in @xmath130 will just affect the absolute value of @xmath131 ( and , consequently , of @xmath112 ) but will not change the features of the forthcoming numerical analysis whose purpose is just to corroborate the analytical results .
a final comment on neutrino masses is appropriate .
if neutrino masses are in the range of the @xmath132 ( i.e. smaller than the temperature at @xmath133 ) , they will be non - relativistic today .
however , they will be relativistic around @xmath133 and will then contribute to the radiation fraction @xmath134 ( and not to the matter fraction ) . in the following ( when writing the evolution equations of the neutrino component ) neutrino masses
will then be neglected since our analysis applies for @xmath135 . in setting the initial conditions ( i.e. for @xmath114 ) the neutrinos are effectively massless since the temperature of the radiation plasma is higher than @xmath136 . for @xmath114 ( but always after neutrino decoupling , i.e. @xmath137 ) the radiation component dominates and , within the gauge - invariant approach presented in section 2 , it is particularly simple to set adiabatic or isocurvature initial conditions and to solve the equations deep inside the radiation - dominated phase .
few relevant cases will now be illustrated in view of the numerical analysis reported in section 3 .
when the plasma is formed by baryons , photons , cdm particles and massless neutrinos , the gauge - invariant density perturbations of each fluid obey @xmath138 where the subscripts @xmath139 , @xmath140 , @xmath141 and @xmath142 refer , respectively , to the baryon , photon , cdm and neutrino components of the plasma . since the divergences of the peculiar velocities carry a spatial gradient , they are small for modes that are larger than the hubble radius prior to equality .
adiabatic initial conditions amount then to the choice that all the density contrasts are equal and constant , i.e. @xmath143 equation ( [ adincon ] ) implies that all the non adiabatic pressure density variation of the fluid vanishes exactly by virtue of eq .
( [ dpnad ] ) . if the initial conditions for the gauge - invariant density contrasts are different from eq .
( [ adincon ] ) , then the corresponding solution is said to be non - adiabatic . in fact , any deviation from the initial conditions ( [ adincon ] ) necessarily leads to a non - vanishing entropy perturbation .
for instance , if @xmath144 , @xmath145 .
this is the situation of the so - called neutrino isocurvature mode .
yet another example could be the one of the cdm - radiation mode where @xmath146 and @xmath147 .
the evolution equations of the velocity fields of the cdm and of the neutrinos are that appear at the right hand side of eqs . ( [ bgamma ] ) and ( [ cnu ] )
are : @xmath148 = 0 , \label{cvel}\\ & & \theta_{\nu } ' - \nabla^2 \biggl [ \biggl(\frac{\psi}{{\cal h}}\biggr ) ' + \frac{{\cal h}^2 - { \cal h}'}{{\cal h}^2 } { \cal r}\biggr ] + \nabla^2 \zeta_{\nu } - \nabla^2 \sigma_{\nu}=0 . \label{nuvel}\end{aligned}\ ] ] the quantity @xmath149 appearing in eq .
( [ nuvel ] ) is the neutrino anisotropic stress .
in fact , since the neutrinos are collisionless particles , their anisotropic stress ( connected to the quadrupole moment of the neutrino phase space distribution as @xmath150 ) is not erased after neutrino decoupling occurring , approximately , when the plasma was dominated by radiation and for temperatures of the order of @xmath151 mev .
the total anisotropic stress introduced in eq .
( [ tfree ] ) will then be written , for convenience , as @xmath152 where @xmath153 is the anisotropic stress of the magnetic fields .
recalling the explicit form of @xmath154 given in eq .
( [ ansb ] ) , it is easy to deduce that @xmath155}{a^4 \rho_{\gamma } } + \frac{1}{4 } \nabla^2 \omega_{\rm b } , \label{anis2}\ ] ] where it has been used that @xmath156 and that @xmath157 . in eq .
( [ anis2 ] ) the ratio bewteen the magnetic energy density and the photon energy density : @xmath158 since @xmath159 , eq .
( [ omb ] ) implies that @xmath160 is constant in time .
therefore , according to eqs .
( [ anis2 ] ) and ( [ omb ] ) , the contribution of the magnetohydrodynamical energy - momentum tensor to the scalar fluctuations reduces , effectively , to the magnetic energy ( and pressure ) densities ( i.e. @xmath161 and @xmath162 ) and to the divergence of the lorentz force .
the collisionless nature of the neutrinos implies that not only the quadrupole moment ( i.e. the anisotropic stress ) of the neutrino phase space distribution contributes to the fluid evolution but also the higher moments should be taken into account .
in particular , the following gauge - invariant equations have to be introduced @xmath163 .
\label{hierarchy}\end{aligned}\ ] ] equation ( [ hierarchy ] ) applies for multipoles larger than the quadrupole . to specify initial conditions for neutrinos
would then require , in principle , all the multipole moments of the ( perturbed ) neutrino phase space distribution . in practice
, it will be sufficient to analyse the initial conditions of the lower moltipoles i.e. monopole ( density constrast ) , dipole ( peculiar velocity ) and quadrupole ( ansotropic stress ) .
unlike the cdm and neutrinos , the evolution of the baryons and of the photons is coupled by thompson scattering : @xmath164 - \nabla^2 \zeta_{\gamma } + a n_{\rm e } x_{\rm e } \sigma_{\rm t } ( \theta_{\rm b } - \theta_{\gamma } ) , \label{photv}\\ & & \theta_{\rm b } ' + { \cal h } \theta_{\rm b } = \nabla^2 \biggl [ \frac{\psi'}{{\cal h } } + \frac{{\cal h}^2 - { \cal h}'}{{\cal h}^2 } ( { \cal r } + \psi)\biggr ] + \frac{\vec{\nabla } \cdot [ \vec{j } \times \vec{b}]}{\rho_{\rm b } a^4 } + \frac{4}{3 } \frac{\rho_{\gamma}}{\rho_{\rm b } } a n_{\rm e } x_{\rm e } \sigma_{\rm t } ( \theta_{\gamma } - \theta_{\rm b } ) , \label{barv}\end{aligned}\ ] ] where @xmath165 is the thompson scattering cross section , @xmath166 is the electron density and @xmath167 is the ionization fraction . by taking the difference of eqs .
( [ photv ] ) and ( [ barv ] ) the evolution of the combination @xmath168 can be obtained with the result that @xmath169 ( \theta_{\gamma } - \theta_{\rm b})= { \cal h } \theta_{\rm b}- \nabla^2(\zeta_{\gamma}+\psi ) - \frac{\vec{\nabla } \cdot [ \vec{j } \times \vec{b}]}{\rho_{\rm b } a^4}. \label{tc1}\ ] ] for @xmath114 the characteristic time scale for the synchronization of photons and baryons is set by @xmath170 which is small in comparison both with the expansion time , i.e. @xmath171 , and with the typical oscillation time , i.e. @xmath172 where @xmath173 is the fourier wavenumber .
consequently , any deviation of @xmath168 rapidly decays away .
this aspect can be appreciated from eq .
( [ tc1 ] ) by taking the limit @xmath174 .
no matter how strong the forcing term is , @xmath175 .
because of the tight coupling between photons and baryons , the baryon - photon velocity as a single physical quantity , i.e. @xmath176 whose evolution equation can be obtained by multiplying eq .
( [ photv ] ) by @xmath177 and by adding it to eq .
( [ barv ] ) ( multiplied by @xmath178 ) .
the scattering terms then cancel and the evolution of @xmath176 , becomes : @xmath179- \frac{\nabla^2(\psi+ \zeta_{\gamma})}{1 + r_{\rm b } } + \frac { 4 \nabla^2 \sigma_{\rm b } - \nabla^2\omega_{\rm b } } { 4 ( 1 + r_{\rm b } ) } .
\label{tc}\end{aligned}\ ] ] to derive the last term in eq .
( [ tc ] ) , eq .
( [ anis2 ] ) has been used to express the lorentz force in terms of the anisotropic stress and of the magnetic energy density in units of the photon energy density . in eq .
( [ tc ] ) the baryon to photon ratio has been also introduced as @xmath180 where @xmath181 is the redshift with the conventions adopted in eq .
( [ ai nt ] ) .
( [ tc ] ) involves only gauge - invariant quantities , it is simple to obtain each corresponding equation in a specific gauge .
for instance , the conformally newtonian gauge is often employed for numerical discussions of the tight - coupling approximation . in this gauge the perturbed metric components as simply @xmath182 and @xmath183 .
the gauge - invariant quantities appearing in eq .
( [ tc ] ) read , in the longitudinal gauge , @xmath184 where @xmath185 and @xmath186 are the density contrasts computed in the conformally newtonian gauge .
equation ( [ tc ] ) becomes , after easy algebra , @xmath187 which is the standard form to the tight coupling equations in the conformally newtonian gauge if @xmath188 .
it is useful to mention that the total @xmath13 ( see eq .
( [ zetasum ] ) ) is given by @xmath189 where @xmath190 , @xmath191 , @xmath178 and @xmath192 are , respectively , the unperturbed energy densities of neutrinos , photons , baryons and cdm particles .
this observation is relevant for the actual solution of the hamiltonian constraint ( [ ham1 ] ) . finally , to solve the momentum constraint ( [ mom0 ] ) , we recall that the total peculiar velocity @xmath78 is given , in terms of the peculiar velocities of each fluid , by the following expression @xmath193 with the same notations of eq .
( [ zetatt ] ) , @xmath194 , @xmath195 while @xmath196 . by going to fourier space , eqs .
( [ bgamma ] ) and ( [ cnu ] ) can be easily solved , to lowest order in @xmath197 and for @xmath114 . denoting with @xmath198 the curvature perturbation on super - hubble scales and with @xmath199 value of the bardeen potential in the same epoch we will have that the hamiltonian and momentum constraints imply @xmath200 in eq .
( [ ad3 ] ) the fractions of neutrinos and photons present in the radiation - dominated plasma have been introduced according to the following conventions : @xmath201 with @xmath202 , @xmath203 .
the gauge - invariant density contrasts will then be determined to be @xmath204 by explicit integration of the corresponding equations , the expressions of the peculiar velocities are @xmath205 , \label{ad5}\\ & & \theta_{\gamma\,{\rm b } } = - k^2 \tau \biggl[(\psi _ { * } + { \cal r } _ { * } ) + \sigma_{\rm b } - \frac{r_{\nu}}{4}\omega_{\rm b } \biggr ] .
\label{ad6}\end{aligned}\ ] ] finally , the expression of the neutrino anisotropic stress turns out to be : @xmath206 from eq .
( [ ad3 ] ) , @xmath199 is determined from @xmath198 and from the parameters of the magnetized plasma , therefore the @xmath199 appearing in the other formulae have been used just to avoid too cumbersome expressions . in the limit @xmath207 and
@xmath208 the solution reproduces the standard adiabatic mode .
using eqs .
( [ t1 ] ) and ( [ t2 ] ) relating the gauge - invariant quantities with the conformally newtonian fluctuations , we obtain , for instance @xmath209 if @xmath207 and @xmath208 , eq .
reproduces the well known result with the difference of the newtonian potentials being constant ( to zeroth order in @xmath210 ) and proportional to the fraction of neutrinos @xcite .
the discussion presented in the case of the adiabatic mode can be repeated , with the due differences , in the case of the isocurvature modes . here
only few examples will be given with particular attention to those features that will be useful for the forthcoming numerical aspects of the analysis .
in the case of the neutrino isocurvature density mode , to lowest order in @xmath210 , the bardeen potential is given by the following expression ; the value of @xmath199 is determined by @xmath211 according to eq .
( [ nu1 ] ) . for practical purposes ,
however , we will give some of the quantities as a function of both @xmath211 and @xmath199 . ]
@xmath212 , \label{nu1}\ ] ] where the term @xmath211 is the specific entropy of the photon - neutrino system .
the term @xmath211 is given , in the present approach , exactly by the mismatch between the density contrasts of the photons and of the neutrinos , i.e. @xmath213 .
the density contrasts of each individual species are then : @xmath214 again , integrating the evolution equations of the peculiar velocities the following results can be obtained : @xmath215 , \label{nu5}\\ & & \theta_{\gamma{\rm b } } = - k^2 \tau \biggl [ 2 \psi _ { * } + \frac{r_{\nu}}{3 } { \cal s } _ { * } - \frac{r_{\gamma } + 1}{4 } \omega_{\rm b } + \sigma_{\rm b } \biggr ] . \label{nu6}\end{aligned}\ ] ] a feature of the neutrino isocurvature density mode is that the total density constrast , @xmath13 vanishes to zeroth order in @xmath216 .
consequently also the total curvature fluctuation , i.e. @xmath27 , vanishes in for typical wavelengths larger than the hubble radius .
the neutrino anisotropic stress is given , in this case , by @xmath217 it is useful to make contact with the conformally newtonian gauge where , according to eqs .
( [ t1 ] ) and ( [ t2 ] ) , the two newtonian potentials are @xmath218,\qquad \psi _ { * } = - \frac { 4 r_{\gamma } r_{\nu}}{3 ( 4 r_{\nu } + 15 ) } \biggl [ s _ { * } - \frac{3}{r_{\nu } } \sigma_{\rm b } + \frac{3}{4 } \omega_{\rm b } \biggr ] .
\label{nu8}\ ] ] notice that @xmath219 .
this implies , in the conformally newtonian gauge , that @xmath220 vanishes , as already pointed out by looking directly at the gauge - invariant fluctuations @xmath13 and @xmath27 .
consider finally the situation where the non vanishing fluctuations in the specific entropy are given by @xmath221 and @xmath222 and define , conventionally , @xmath223 this is the case of the cdm - radiation isocurvature mode .
then , from eq .
( [ dpnad ] ) , by making the sum over the fluid components explicit we do obtain @xmath224 where we used that @xmath225 . now defining @xmath226 , eq .
( [ cdm2 ] ) can also be written as @xmath227 where the second equality in the definition of the total sound speed @xmath228 follows from the conventions illustrated in connection with eq .
( [ ai nt ] ) .
therefore , to lowest order in @xmath229 and for @xmath230 we have that @xmath231 where we used that @xmath232 ( so that @xmath233 ) . in the limit @xmath234 , eq . ( [ cdm4 ] ) implies that @xmath235 which reproduces the standard result obtainable for the cdm - radiation isocurvature mode in the absence of magnetic fields . in the same limit of vanishing magnetic contribution ( and with vanishing fractional contribution of neutrinos , i.e. @xmath236 ) @xmath237 which
is also well known from the analysis of the two fluid system in the absence of neutrinos . in this case
the relevant density contrasts are given by @xmath238 as it follows from the hamiltonian constraint and from the definition of @xmath13 in terms of the density contrasts of each fluid .
the peculiar velocities can then be directly obtained to lowest order . in particular
it is interesting to comment about the neutrino sector where the solution for the lowest multipoles can be written as @xmath239 k^2 \tau^2 + k\tau \biggl [ \frac{r_{\gamma}}{r_{\nu } } \sigma_{\rm b } - \frac{\omega_{\rm b}}{4 } r_{\gamma } \biggr ] , \nonumber\\ & & \sigma_{\nu } = - \frac{r_{\gamma}}{r_{\nu } } \sigma_{\rm b } + \frac{k^2 \tau^3}{6 } \frac{\overline{\omega}_{\rm c}}{15 + 2 r_{\nu } } \biggl [ s _ { * } + \frac{3}{4 } \omega_{\rm b } r_{\gamma } \biggr ] , \nonumber\\ & & { \cal f}_{\nu\,3 } = \frac{8}{15 } \biggl [ \frac{r_{\gamma}}{r_{\nu } } \sigma_{\rm b } - \frac{\omega_{\rm b}}{4}\biggr ] k\tau . \label{cdm7}\end{aligned}\ ] ] notice that to set correctly initial conditions in the magnetized case the octupole is important .
however , in the limit of vanishing magnetic inhomogeneities , the octupole can be simply set to zero as all the remaining terms in the hierarchy .
finally , in the case of the cdm - radiation mode , using eqs .
( [ t1 ] ) and ( [ t2 ] ) , the conformally newtonian potentials are : @xmath240 the two longitudinal potentials @xmath241 and @xmath81 are only equal in the limit @xmath242 . in the two - fluid treatment of the problem
this would actually be the case since no neutrinos are allowed .
the regular solutions describing both adiabatic and entropy modes will now be followed through the transition between the radiation - dominated stage of expansion and the matter - dominated epoch .
in fact , as already mentioned , the decoupling time takes place after @xmath112 ( i.e. @xmath243 for @xmath244 ) . in the absence of magnetic fields and in the absence of non - adiabatic pressure density fluctuations , eqs .
( [ ham1 ] ) and ( [ zetamas ] ) imply that the total curvature perturbation @xmath27 , i.e. the curvature fluctuation on comoving orthogonal hypersurfaces , is conserved for wavelengths much larger than the hubble radius , i.e. for @xmath216 .
conversely , the bardeen potential @xmath29 diminishes , in the transition from radiation to matter of a factor @xmath245 ( assuming vanishing neutrino anisotropic stress ) . in the absence of magnetic fields , the effect of the neutrino anisotropic stress is often ignored for order of magnitude estimates . however , if the plasma is magnetized , this approximation is no longer justified as clearly demonstrated by working out the solutions presented in section 3 .
in fact , the anisotropic stress of the neutrinos is balanced by the magnetic anisotropic stress and the subsequent evolution of the system will tell what happens once the initial conditions are accurately set .
adiabatic ( magnetized ) initial conditions are imposed according to eqs .
( [ ad1 ] ) , ( [ ad2 ] ) and ( [ ad3 ] ) . the initial conditions for the density contrasts can be found in eq .
( [ ad4 ] ) .
the velocity fields and the neutrino anisotropic stress are then the ones derived , respectively , in eqs .
( [ ad5])([ad6 ] ) and ( [ ad7 ] ) . by setting @xmath246 in the initial conditions the standard adiabatic mode ( with canonical dependence upon the fractional contribution of massless neutrinos )
is reproduced . in the realistic situation
the adiabatic mode will have a nearly scale - invariant spectrum ( with red tilt ) while @xmath247 will have a nearly scale - invariant spectrum ( with blue tilt ) . for simplicity
the relevant results concerning the power spectra are collected in the appendix here we just recall that the spectra of @xmath198 and @xmath247 and @xmath153 can be written , according with the notations of the appendix , as @xmath248 and as @xmath249 where @xmath250 is a conventional pivot scale of the scalar adiabatic modes and @xmath251 is the pivot scale usually employed to parametrize the magnetic power spectra .
the quantity @xmath252 is the spectral tilt of the magnetic energy spectrum i.e. @xmath253 . through the years it emerged that a nearly scale invariant magnetic energy spectrum is probably the most interesting to discuss in cmb studies
since it is not strongly affected by diffusive processes ( see , for instance , @xcite ) .
the other numerical quantities appearing in eq .
( [ sp1 ] ) are defined in eqs .
( [ feps ] ) and ( [ omc2 ] ) of the appendix .
if the magnetic fields are absent and @xmath254 , the sachs - wolfe plateau ( see section 5 ) corresponds to @xmath255 the first interesting exercise ( even if not realistic as we shall see later on ) is to assume that the magnetic field is exactly comparable in magnitude with the curvature perturbations , i.e. @xmath256 . in particular without loss of generality , we will assume @xmath257 and @xmath258 .
in this situation , as dictated by eq .
( [ ad7 ] ) total anisotropic stress @xmath259 is initially vanishing for typical wavelengths larger than the hubble radius . figure [ f1 ] ( plot at the left hand side ) shows that the total anisotropic stress remains vanishing as long as the mode is outside the horizon and it oscillates as soon as the mode reenters . in fig .
[ f1 ] the time around the radiation - matter transition is measured in @xmath260 while the comoving wave - numbers are measured in @xmath261 . in the right plot of fig .
[ f1 ] we use the same initial conditions but with a value of @xmath247 that is one tenth of the amplitude of the initial curvature perturbations @xmath198 : the amplitude of the oscillations of the total anisotropic stress diminishes . [
cols="^,^ " , ] in order to discuss this situation in fully consistent terms it is necessary to perform a numerical analysis of parameters extraction similar to the one reported in @xcite where , however , the magnetized contribution is taken into account both at the level of the initial conditions and at the level of the normalization . to get an estimate of the possible bounds obtainable in the present framework on the intensity of the magnetic field the minimal situation
will be considered .
suppose , indeed , that the sw plateau is just produced by a magnetized adiabatic mode . in this case eq . ( [ cell2 ] ) reduces to @xmath262 .
\label{cell3}\ ] ] where the correlation angle has been simply denoted by @xmath140 , since it is , in this case , the only correlation angle .
since the contribution to the sw plateau must be predominantly adiabatic we do know that the scalar spectral index must be nearly scale invariant . in the case where @xmath257 and in the absence of magnetic field , eq . ( [ cell3 ] ) implies that @xmath263 .
the wmap determination will then impose @xmath264 .
the recent release of wmap data @xcite seems to prefer slightly red spectra with preferred value ( taking wmap alone ) @xmath265 .
the spectrum of the magnetized contribution is controlled now by @xmath252 . if @xmath266 we are in the interesting situation where also the magnetic field spectrum is nearly scale - invariant ( though with a blue tilt since @xmath252 is directly related with the tilt of the magnetic energy spectrum @xmath267 , in the notations of the appendix ) .
the range @xmath268 is phenomenologically interesting not only because , in this range , the magnetic energy spectrum is nearly scale - invariant but also because the diffusive effects ( i.e. finite thermal and magnetic diffusivity scales ) do not affect the properties of the large - scale magnetic field ( see , for instance , @xcite ) .
if the contribution to the sw plateau is predominantly adiabatic , then , the limit on the magnetic field intensity for a given range of @xmath269 and @xmath270 are illustrated in fig .
[ fb1 ] assuming the validity of eq .
( [ cell3 ] ) and enforcing the wmap ( i.e. experimental ) determination of the sw plateau . in the left plot of fig .
[ fb1 ] the maximal magnetic field intensity is illustrated as a function of the scalar spectral index @xmath269 for different values of @xmath252 . for the selected values of the parameters the magnetic field has to be always below each of the curves .
as discussed above ( and also in the appendix ) , @xmath271 is the magnetic field regularized over a typical comoving wave - number @xmath272 that we take to be , according to the conventions , @xmath151 @xmath261 .
the pivot scale has been fixed at @xmath273 . to obtain the estimates illustrated in fig .
[ fb1 ] we assumed , in eq .
( [ cell3 ] ) the less favourable situation , i.e. the one where @xmath274 . in this case
( maximal antcorrelation ) the two autocorrelations and the cross - correlation term all add - up to the plateau and there is no possibility of ( artificial ) reduction of the amplitude . from the left plot of fig .
[ fb1 ] it is clear that the present value of the magnetic field is bounded to be @xmath275
. similar conclusion can be drawn from the plot at the right hand side of fig .
[ fb1 ] where the allowed region is illustrated for a specific value @xmath269 ( preferred by wmap data alone ) as a function of @xmath252 .
it is clear that by allowing a positive correlation ( i.e. @xmath276 ) instead of a negative one ( as previously assumed ) the bounds may be relaxed .
it could be argued that the limit on magnetic fields arising from this analysis is competitive with the ones obtainable , for instance , from faraday rotation of the e - modes into b - modes .
the main conclusion is however , that it is now possible to include the magnetic fields in a systematic strategy of parameter extraction from available cmb data possibly combined with large scale structure data .
there are no compelling reasons why the scalar modes induced by large - scale magnetic fields should not be considered as a possible source of cmb anisotropies . in this paper a new framework for this analysis has been proposed .
the main findings can be summarized as follows : * the presence of large - scale magnetic fields ( with nearly scale - invariant magnetic energy spectrum ) affects both the adiabatic and the entropy modes ; * a general gauge - invariant system for the analysis of the initial conditions has been proposed : in such a system the gravitational sector is described in terms of the curvature perturbations on comoving orthogonal hypersurfaces , the density constrast on uniform density hyepersurfaces and the bardeen potential ; * this gauge - invariant system has been analytically solved deep in the radiation dominated epoch ( well before equality ) and then numerically followed after equality ( but before recombinantion ) in the tight coupling approximation ; * the numerical estimates ( corroborated by the analytical results ) allow to compute the precise value of the magnetized adiabatic and non - adiabatic curvature perturbations when the relevant modes are outside the horizon before decoupling ; * as an example the effect of magnetic fields on the sw plateau has been discussed .
the magnetized curvature and entropy perturbations propagate then smoothly through equality .
this result seems particularly useful since the common lore was that the presence of magnetic fields induce a singular modes .
as discussed , this presumption was a simple consequence of the fact that , in most studies , the anisotropic stress of the neutrinos was simply neglected . in the present analysis , on the contrary , the four - fluid system describing the most general initial conditions for cmb anisotropies has been consistently studied and solved .
the present paper opens various possible developments .
first of all the analysis reported in the present paper involves the lowest order in the tight coupling expansion
. it will be interesting to go beyond and study , within the same framework , the effects arising to first - order in the tight coupling expansion .
this will allow the treatment of various phenomena ( including faraday effect ) that have been discussed , up to now , just assuming the standard adiabatic mode that is , however , itself modified by the presence of the magnetic field as demonstrated in the present paper .
numerical studies on the relevance of magnetized fluctuations correlated ( or anticorrelated ) with adiabatic and non - adiabatic modes are now conceptually possible since the super - horizon curvature perturbations have been carefully computed ( after equality ) starting from the solution of the system of perturbations after neutrino decoupling .
possible strategies of parameter extraction adapted to this new situation should be discussed in future investigations .
in the present appendix the main conventions for the evaluation of the power spectra will be summarized .
this technical discussion is relevant since there are conventions that are used in the magnetic field literature ( for instance concerning the relative normalization of the power spectra ) and that differ from the conventions used in cmb physics .
according to the conventions adopted in this paper , the fourier components of the curvature and entropy fluctuations are defined as @xmath277 where @xmath278 and @xmath279 .
the correlation functions of the primordial adiabatic and isocurvature modes are then defined as @xmath280 where @xmath281 and where the power spectra can be parametrized as @xmath282 in eq .
( [ psrs ] ) , @xmath283 is the pivot " scale at which the spectra are normalized .
a typical choice is @xmath284 @xmath261 .
see @xcite and also @xcite for a discussion of the sensitivity of the parameter estimation on @xmath283 . ] for @xmath283 is @xmath285 .
if we allow a correlation between adiabatic and isocurvature modes , the initial value problem may require a further primordial correlation function , namely : @xmath286 where @xmath287 is the correlation angle . if @xmath288 the primordial modes are not correlated . with similar notations , the fourier modes for the magnetic field intensity are @xmath289 again with @xmath290 .
the magnetic fields studied in the present paper are fully inhomogeneous and characterized by their two - point function , i.e. @xmath291 where @xmath292 the amplitude of the magnetic power spectrum , @xmath293 , can be traded for the magnetic energy density regularized over a typical scale of size @xmath294 .
in fact , by using a gaussian window function @xmath295 for each magnetic field intensity , the ( real space ) magnetic autocorrelation function at two coincident spatial points is @xmath296 where @xmath297 .
equation ( [ ab ] ) implies that the amplitude of the magnetic power spectrum appearing in eqs .
( [ bcorr ] ) and ( [ mps ] ) can be expressed in terms of @xmath298 as @xmath299 where , for convenience , the magnetic spectral tilt @xmath300 has been introduced .
different authors ( see , for instance , @xcite ) employ slightly different conventions both for eqs .
( [ mps])([atob ] ) .
in particular , sometimes the factor @xmath301 appearing in eq .
( [ bcorr ] ) is dropped and reabsorbed in the definition of @xmath302 which is , in turn , often parametrized as @xmath303 ( without any typical pivot scale ) .
if these two choices are consistently adopted , a nearly scale invariant magnetic energy spectrum ( see below ) occurs for @xmath304 . within the present conventions , that are homogeneous with the ones customarily adopted in the analysis of primordial adiabatic and isocurvature modes , the nearly scale - invariant magnetic energy spectrum is obtained in the range @xmath305 .
the nearly scale - invariant spectrum is particularly interesting for the purposes of the present investigation . in this limit , in fact ,
the magnetic field energy spectrum ( as well as the spectrum of the lorentz force ) is practically unaffected by diffusive damping provided by the effective alfvn damping scale @xcite .
the alfvn damping occcurs for typical wavelengths that are of the order of @xmath306 where @xmath307 is the effctive alfvn velocity and @xmath308 is the silk damping scale . since @xmath307 will be proportional , in the present notations , to @xmath309 , we can also appreciate that @xmath310 will simply be a fraction of the silk scale . in the case covered by the present investigation , on the contrary
, we are not interested in setting bounds on very steep ( violet ) spectra of the magnetic energy density that are , in any case , already strongly constrained by the analysis of the vector and tensor modes . on the contrary , since in the present analysis , the magnetic fields induce some correlated isocurvature modes , it is also reasonable to deepen the analysis in the nearly scale - invariant case . with these specifications , the typical scale where diffusion becomes important will not be directly related with the alfvn damping but rather with the typical scale at which the wavelength of the fluctuations becomes comparable with the photon mean free path corresponding to a comoving wave - number @xmath311 from eq .
( [ bcorr ] ) it is possible to compute , by convolution , the appropriate quadratic observables appearing in the treatment of the scalar problem . in particular , as discussed in the bulk of the paper , the effect of the magnetic fields on scalar modes depends upon the magnetic energy and pressure densities , ( i.e. @xmath161 and @xmath312 ) as well as on the divergence of the lorentz force .
this last quantity , in mhd , is proportional to @xmath313 $ ] .
recalling , in fact , the notations established in eqs .
( [ anis2 ] ) and ( [ omb ] ) we have that , in real space @xmath314 where @xmath315 in eq .
( [ omega ] ) the quantity @xmath316 has been introduced for practical reasons . consistently with eq .
( [ defft2 ] ) , the correlation function of the magnetic energy density is then : @xmath317 where @xmath318 notice that the obtained expression for @xmath267 follows from the mode - coupling integral @xmath319 ^
2}{k^2 ( \vec{q } + \vec{k})^2 } \biggr\ } p_{\rm b } ( \vec{q } + \vec{k } ) p_{\rm b } ( \vec{k } ) .
\label{omcorr}\ ] ] in the limit @xmath320 .
according to eq .
( [ anis2 ] ) , the correlator of @xmath153 determines , together with @xmath247 the lorentz force .
as far as the two - point function of @xmath153 is concerned , we can start by noticing that @xmath321 where @xmath322.\ ] ] consequently , @xmath323 where , for @xmath324 , @xmath325 in the general phenomenological situation , the magnetic field may also be correlated with the primordial adiabatic and isocurvature modes . in this case ,
two new cross - correlations have to be taken into account when computing , for instance , the sachs - wolfe plateau : @xmath326 again , in eq .
( [ crossc ] ) the various angles " parametrize the correlation between different modes . if @xmath327 the modes are not correlated
|
an appropriate gauge - invariant framework for the treatment of magnetized curvature and entropy modes is developed .
it is shown that large - scale magnetic fields , present after neutrino decoupling , affect curvature and entropy perturbations .
the evolution of different magnetized modes is then studied across the matter - radiation transition both analytically and numerically . from the observation
that , after equality ( but before decoupling ) the ( scalar ) sachs - wolfe contribution must be ( predominantly ) adiabatic , constraints on the magnetic power spectra are deduced .
the present results motivate the experimental analysis of more general initial conditions of cmb anisotropies ( i.e. mixtures of magnetized adiabatic and isocurvature modes during the pre - decoupling phase ) .
the role of the possible correlations between the different components of the fluctuations is partially discussed . 8.5 in 6.5 in -30pt 0.2 cm cern - ph - th/2006 - 056 +
|
it is a prediction of the concordance cold - dark - matter cosmological model that galaxies and galaxy clusters are surrounded by a massive dark matter ( dm ) haloes .
firm evidence for the dm existence is coming from various gravitational effects in astrophysics and cosmology @xcite .
if the existing cosmological dm @xcite is a thermal relic consisting of weakly interacting massive particles , dm annihilations into the standard model particles should provide indirect evidence of dm in cosmic ray experiments @xcite .
unfortunately the direct @xcite and indirect @xcite searches for dm particles have all given either negative or contradictory results .
a notable exception to this result is the recent evidence for @xmath1-ray excess with energy 130 gev @xcite in the fermi large area telescope ( lat ) @xcite data .
this excess originates predominately from a small region in the galactic centre @xcite and may have a double peak structure @xcite .
its global statistical significance is between @xmath2 and @xmath3 @xcite , depending whether one or two peaks are fitted , and it is consistent with the fermi - lat bound on monochromatic photon lines from diffuse @xmath1-ray data @xcite .
although galactic background effects make statistical fluctuation in the @xmath1-ray spectrum more likely than naively expected @xcite , the presence of a double peak in the background is very unlikely .
although an option that the 130 gev @xmath1-ray excess is a fake detector effect is not entirely excluded , recent studies disfavour this possibility @xcite .
it is likely that fermi - lat has either observed an astrophysical process that unexpectedly gives photon peak(s ) or detected the dm annihilations into monochromatic photons . to verify that the dm has been discovered indirectly ,
the @xmath1-ray excess must either be confirmed by other experiments such as the planned high - resolution experiments or , or to observe the same excess with fermi - lat from other known dm dominated objects .
the expected signal from nearby dwarf galaxies turned out to be too weak to check the 130 gev @xmath1-ray excess with fermi - lat @xcite .
however , the galaxy clusters , the biggest nearby cosmological structures dominated by dm , are expected to be much better objects for that purpose @xcite because the dm annihilation signal from there should be amplified by a boost factor due to the existence of many dm subhaloes @xcite .
there are large uncertainties in theoretical predictions of the boost factors @xcite , numerical estimates vary from 10 to 10000 .
experimental measurements are needed to discriminate between the different subhalo models . here
we report on searches for spectral features in fermi - lat @xmath1-rays from regions corresponding to nearby galaxy clusters determined by the magnitude of their signal line - of - sight integrals ( @xmath4-factors ) .
we observe a double peak - like excess at photon energies 110 gev and 130 gev over the diffuse power - law background with statistical significance up to @xmath0 , confirming independently the earlier claims of excess from the galactic centre . interpreting this result as a signal of dm annihilations into two channels with monochromatic final - state photons , and fixing the annihilation cross section from galactic centre data ,
we determine the annihilation boost factor due to galaxy cluster subhaloes .
in this work we search for spectral features in the @xmath1-ray spectrum originating from the known galaxy clusters in fermi - lat data . for that , we work with a set of galaxy clusters which parameters are reliably known in the literature .
we compute their @xmath4-factors according to @xcite that allows us to select clusters according to their expected contribution to the dm annihilation signal .
coordinates , masses , distances and radii for the 18 galaxy clusters with largest @xmath4-factors are collected in table [ tab1 ] . because of limited statistics we can not study signal from every cluster individually .
instead we sum up all the photons coming from the directions of these clusters and study the spectrum of stacked flux .
current determinations of the galaxy cluster dm halo parameters suffer from large uncertainties . despite of
that we do observe a correlation between the @xmath4-factors of the galaxy clusters and the number of signal photons from them ( see fig .
[ fig : jfactor ] ) , supporting our claim that the signal originates from galaxy clusters .
we reanalyse the galactic centre excess with new lat resolution and find the double peak in the same position , supporting our claim that the two spectra both signal dm annihilations . assuming this , we determine the dm annihilation boost factor in galaxy clusters from fermi lat data . in the present analysis we consider the public fermi - lat photon event data of 218 weeks within energy region from 20 to 300 gev .
we apply the quality - filter cuts , as recommended by the fermi - lat team , as we did in our previous study @xcite .
we make use of the events selection ( pass 7 version 6 ) , in order to minimise potential systematical errors .
the selection of events as well as the calculation of exposure maps is performed using the fermi .
the most important improvement compared to our previous work @xcite is the usage of new improved fermi - lat energy resolution @xcite . to avoid effects of point sources we exclude photons that are within an energy - independent cut radius of each source .
we used all 1873 sources from the lat 2-year point source catalog @xcite .
the cut radius is chosen to be 0.2@xmath5 @xcite .
we also tested the radii 0@xmath5 , 0.15@xmath5 , 0.25@xmath5 and 0.5@xmath5 and find no effect on the final result .
to estimate the photon spectrum related to the clusters we select photons that are within an energy independent radii around the centre of each cluster
. we will denote it as the region of interest ( roi ) below .
the annihilation signal should arise predominantly within @xmath6 in each cluster .
the boosting effect due to halo substructures should make the signal spatially flat @xcite .
the morphology of expected signal from a single main halo without substructure should be very different : it should arise from a small central region and must have very cuspy nature @xcite . however , in order to determine the stability signal significance over background , we need to consider larger and different regions than @xmath7 around clusters .
we studied both the radii @xmath7 dependent and independent rois around the clusters .
we found that the result is independent of how the background photons are included signal peaks are not affected while the background fluctuates within estimated errors .
thus the simplest choice of roi is the equal radii for all clusters .
we considered radii @xmath8 3 , 4 , 5 , 6 and 8 degrees for roi and found that starting from smaller values of @xmath9 , at @xmath10 the significance reached maximal value and remained stable for larger radii .
we also found that galaxy clusters with small @xmath4-factors mostly contribute to the background reducing the signal background ratio .
starting from smaller number of clusters , the significance reached the maximum at 5@xmath117 clusters and remained stable beyond that . for numerical results
we therefore consider only the 18 most relevant galaxy clusters presented in table [ tab1 ] . to compute the @xmath1-spectrum we sum up all photons from the selected cluster regions .
the spectrum is calculated by the kernel smoothing as described in detail by @xcite .
the characteristic kernel size is chosen based on the fermi - lat energy respond function .
we calculated the spectrum in logarithmic and linear energy scales and used different kernel functions and sizes .
the results are rather insensitive to the exact kernel function and size , showing that this kernel smoothing method is rather robust . to estimate the signal significance we select @xmath12 random cluster - size regions in sky and find the @xmath1-spectrum for them . to avoid the crowded region at the galactic plane and centre of the galaxy we exclude the region @xmath13 from the study @xcite .
it means the border of a randomly selected region can not be closer to the galactic plane than @xmath14 .
we tested different sizes of the excluded regions : @xmath15 5 , 10 and 15 degrees having neglectable effect on results . for significance estimation of the spectral features we repeat the procedure 100,000 times for all the selected cases of radii to get the distribution of spectra .
based on the 100,000 monte carlo realisations we estimated the confidence limits of the spectra . to estimate the boost factor
, we use the relation between the number of signal photons @xmath16 within solid angle @xmath17 and the properties of dm particle : @xmath18 where @xmath19 is the exposure of cluster region @xmath17 , @xmath20 is the averaged cross - section , @xmath21 is the mass of dm particle and @xmath22 is the number of produced photons per annihilation ( in case of non - self - conjugated particle there is addition factor two in front of @xmath21 ) .
the @xmath4-factor @xmath23 is defined by the line - of - sight integral @xmath24 where @xmath25 is the density profile of dm in galaxy cluster .
the parameters of the main dm halo is considered from @xcite .
we take the annihilation cross - section to photons to be @xmath26 , where @xmath27 @xmath28 s@xmath29 is the standard thermal cross - section , as computed from the @xmath1-ray line signal from the galactic centre @xcite .
fig . [ fig1 ] shows the measured spectrum of @xmath30 galaxy clusters for two choices of fixed radii , @xmath31 and @xmath32 together with the 95% cl background estimated with monte carlo ( grey band ) . reduced signal from the galactic centre
is also presented for comparison . due to the improved energy resolution ( smaller smoothing kernel )
we observe double peak structure in both spectra .
the first peak is at 110 gev and the second at 130 gev , consistently with two dm annihilation channels to @xmath33 and @xmath34 respectively .
this is consistent with a generic prediction of gauge theories .
looking for the double - peak excess with at least the same strenght as observed lines ( 110 gev and 130 gev ) in the whole mc sample and energy region ( 20 300 gev ) implies the global significance of the double - peak signal with @xmath0 .
table [ tab2 ] presents the number of photons in the signal region divided into 5 gev energy bins for different angular regions around the galaxy clusters .
we see that the photons are clustered around two energies ( 110 and 130 gev ) corresponding to the two peaks in fig .
[ fig1 ] . for larger values of considered
radii the statistics increases , giving an optimal signal - over - background ratio for @xmath10 regions . for larger radii , such as @xmath35
, we observe slight reduction of the signal significance , probably due to a fact the background starts dominating and it is not stable for so large roi . assuming that the signal originates from dm annihilation
, we estimate the boost factor due to dm substructures in the galaxy cluster . if this assumption is correct , this is the first real measurement of the boost factor from experimental data . considering the radius @xmath36 ,
the boost factor of substructure turns out to be @xmath37 at 68% cl , where the variation is estimated with bootstrap analysis .
the error are very large and we note that this error is only statistical . in addition , other sources of error , the distance and mass of clusters , averaging over the selected clusters , the main dm profile of the clusters etc . , are not taken into account .
in addition , the measured value of dm annihilation cross section to the photon channels in the galactic centre , @xmath26 , depends on uncertainties of the galactic dm halo and it has accuracy in order of magnitude . taking into account all those uncertainties , the present experimental precision does not allow to determine the boost factor numerically with a meaningful statistical significance . in order to distinguish between different dm subhalo models ,
new more precise measurements are needed .
we stress that the presence of double - peak feature in fermi lat data are not affected by this uncertainties .
the galaxy clusters , as well as the galactic centre , are fixed objects dominated by dm , thus our result does not suffer from statistical fluctuations related to scanning and choosing arbitrary regions of the sky , nor from possible astrophysical effects from the galactic disc .
the fact that the two signals from unrelated regions of sky , from the galactic centre and from the locations of galaxy clusters , give a double peak - like excesses that precisely coincide ( see fig . [ fig1 ] ) suggests that this is not a statistical fluctuation .
this result disfavours astrophysical explanation to the excess since astrophysical processes should not be exactly the same in galaxy clusters and in the galactic centre .
our result implies that most plausibly the dm of universe has been discovered via indirect detection by fermi - lat .
we have found that fermi - lat data shows a double peak - like excess of @xmath1-rays with energies 110 gev and 130 gev from the nearby galaxy clusters .
the maximal global significance of the excess is @xmath0 .
our result provides an _ independent _ confirmation of the previously claimed @xmath1-ray excess from the galactic centre and supports the interpretation of the excess due to dm annihilations into two monochromatic photon channels . making this assumption , and fixing the dm annihilation cross section from the galactic centre data
, we found the boost factor for dm annihilations due to dm substructures in the galaxy cluster from fermi - lat data .
this is the first measurement of dm boost factor from real data and has potentially important implications for understanding the dm substructures in haloes .
unfortunately the related uncertainties at present are quite large .
more data is needed to discriminate between different theoretical models of dm substructure as well as to discriminate the dm interpretation of the excess from the possible astrophysical origin and systematic detector effects .
we thank l. bergstrom , j. conrad , d. finkbeiner and c. weniger for numerous communications .
this work was supported by the esf grants 8090 , 8499 , 8943 , mtt8 , mtt59 , mtt60 , mjd52 , mjd272 , by the recurrent financing projects sf0690030s09 , sf0060067s08 and by the european union through the european regional development fund .
|
we search for spectral features in fermi - lat gamma - rays coming from regions corresponding to eighteen brightest nearby galaxy clusters determined by the magnitude of their signal line - of - site integrals .
we observe a double peak - like excess over the diffuse power - law background at photon energies 110 gev and 130 gev with the global statistical significance up to @xmath0 , confirming independently earlier claims of the same excess from galactic centre . interpreting this result as a signal of dark matter annihilations to two monochromatic photon channels in galaxy cluster haloes , and fixing the annihilation cross section from the galactic centre data , we determine the annihilation boost factor due to dark matter subhaloes from data .
our results contribute to discrimination of the dark matter annihilations from astrophysical processes and from systematic detector effects as the possible explanations to the fermi - lat excess .
|
acceleration of energetic particles is important for a wide range of astrophysical environments , from stellar magnetospheres , accretion disk / jet systems , supernova remnants and gamma ray bursts to clusters of galaxies .
several mechanisms for particle acceleration have been discussed in the literature which include varying magnetic fields in compact sources , stochastic second order fermi process in turbulent interstellar and intracluster media , and the first order fermi process behind shocks .
an alternative , less explored mechanism so far , involves particle acceleration in magnetic reconnection sites , and this will be the focus of the present work which is a first of a series of papers on this subject . for a comprehensive recent review on particle acceleration mechanisms the reader is referred to @xcite .
magnetic reconnection may occur when two magnetic fluxes of opposite polarity encounter each other . in the presence of finite magnetic resistivity , the converging magnetic lines annihilate at the discontinuity surface and a current sheet forms there .
it is common knowledge that magnetic fields stay frozen in highly conductive fluids .
estimates show that ohmic diffusion for astrophysical scales is absolutely negligible , contributing to the strongly entrenched view that interacting magnetic fields of opposite polarity change their topology at low speeds .
this is determined by the magnetic reconnection speed , and the famous example of two oppositely directed magnetic fluxes in ohmic contact undergoing slow sweet - parker @xcite process of magnetic flux annihilation is frequently invoked . in a more generic context , the magnetic reconnection rate is the speed at which two magnetic flux tubes , which are pushed against each other , can pass through altering the initial field topology .
this is a situation that can be frequently encountered in astrophysical fluids with complex motions . in the sweet - parker model
, it has been shown that particles can accelerate due to the induced electric field in the reconnection zone @xcite.this @xmath0 acceleration process , however , is constrained by the narrow thickness of the acceleration zone which has to be larger than the particle larmor radius and by the strength of the magnetic field .
therefore , the efficiency of this process is rather limited . besides , it also does not predict a power - law spectrum , as generally observed for cosmic rays .
observations have always been suggestive that magnetic reconnection can happen at a high speed in some circumstances , in spite of the theoretical difficulties in explaining it .
for instance , the phenomenon of solar flares suggests that magnetic reconnection should be first slow in order to ensure the accumulation of magnetic flux and then suddenly become fast in order to explain the observed fast release of energy . a model that can naturally explain this and other observational manifestations of magnetic reconnection was proposed by @xcite .
the model appeals to the ubiquitous astrophysical turbulence as a universal trigger and controller of fast reconnection .
the predictions of the model have been successfully tested in numerical simulations by @xcite which confirmed that this speed is of the order of the alfvn speed in the presence of weakly stochastic magnetic field fluctuations .
an important consequence of fast reconnection of turbulent magnetic fields is the formation of a thick volume filled with reconnected small magnetic flux loops .
now , if such turbulent flow is immersed in a current sheet formed by two large scale converging magnetic flux tubes ( such as , e.g. , in the sweet - parker configuration ) , then the three - dimensional magnetic fluctuations or loops will contract and scatter test particles , presenting favorable conditions for energetic particle acceleration in a first order fermi process .
in other words , the particle may bounce back and forth between these converging magnetic mirrors formed by oppositely directed magnetic fluxes moving towards each other with the velocity corresponding to the reconnection speed .
this has been first described in @xcite ( see also * ? ? ?
* ) for the situation when there is no back reaction of the accelerated particles on the reconnected magnetic flux .
later , @xcite appealed to a similar process , but within a collisionless reconnection scenario . in their model ,
the contraction of two - dimensional loops is controlled by the firehose instability that arises in the particle - in - cell ( pic ) simulations containing both electrons and ions @xcite . in the present work ,
we show that this type of acceleration can be present in a pure magnetohydrodynamical scenario as well , where the pressure is fully isotropic and the contraction of islands is determined by their interactions .
magnetic reconnection is ubiquitous in astrophysical circumstances and therefore it is expected to induce acceleration of particles in a wide range of astrophysical environments including galactic and extragalactic ones .
for instance , the process has been already discussed for the production of ultra - high energy cosmic rays @xcite , acceleration of particles in gamma ray bursts @xcite , microquasars @xcite and astrophysical jet - accretion disks in general @xcite . in particular , in the case of relativistic jets , a diagram of the magnetic energy rate released by violent reconnection as a function of the black hole ( bh ) mass spanning @xmath1 orders of magnitude was derived and demonstrates that the magnetic reconnection power is more than sufficient to explain the observed radio outbursts , from microquasars to low luminous active galactic nuclei ( agn ) @xcite .
more recently , the acceleration in reconnection regions has obtained observational support .
it was suggested in @xcite that anomalous cosmic rays measured by voyagers are , in fact , accelerated in the reconnection regions of the magnetopause ( see also * ? ? ?
such a model explains why voyagers did not see any signatures of acceleration passing the solar system termination shock . in a separate development ,
@xcite have appealed to the energetic particle acceleration in the wake produced as the solar system moves through interstellar gas to explain the excess of cosmic rays of the range of both sub - tev and multi - tev energies in the direction of the magnetotail .
note , that due to the 11-year solar magnetic cycle the accumulation of magnetic reversals in the wake is unavoidable . the implications of the acceleration process in reconnection sites are expected to be even much wider .
numerical two - dimensional ( 2d ) simulations recently presented in @xcite confirmed the high efficiency of particle acceleration in regions of magnetic reconnection .
however , we will show here that the process of acceleration happens rather differently in two and three dimensional ( 3d ) situations .
the 3d geometry shows a wider variety of acceleration regimes and this calls for much more detailed studies of the acceleration . in section [ sec :
methods ] we describe the methodology of the particle acceleration studies presented here . in section
[ sec : results ] we show the results obtained from these studies , in sections [ sec : discussion ] and [ sec : summary ] we discuss the results and draw the main conclusions of this work .
in order to study particle acceleration in magnetic reconnection sites , we performed numerical simulations solving the isothermal magnetohydrodynamic ( mhd ) equations in two and three dimensions ( 2d and 3d , respectively ) . to compare our mhd results with those obtained with the particle in cell ( pic ) code in @xcite
we intentionally , in our 2d simulations , have reproduced their set up of eight harris current sheets in a periodic box .
the initial density profile is chosen in such a way that the total ( gas plus magnetic ) pressure is uniform .
initially , we imposed random weak velocity fluctuations to this environment in order to enable spontaneous reconnection events and the development of magnetic islands .
we evolved the system in 2d ( both without and with a guide field along the third dimension defined by z direction ) and in 3d domains . after the development of the magnetic loops ( which in a 2d geometry form islands ) in an underlying plasma configuration with defined density , velocity and magnetic field profiles , we injected test particles at a given snapshot and integrated their trajectories solving the equation of motion for each charged particle @xmath2 where @xmath3 , @xmath4 and @xmath5 are the particle mass , electric charge and velocity , respectively , @xmath6 and @xmath7 are the electric and magnetic fields , respectively , @xmath8 is the lorentz factor , and @xmath9 is the speed of light . in the studies below we assume that the charged particles are protons . in the mhd simulations the electric field @xmath6
is generated either by the magnetized plasma or by resistive effects and can be obtained directly from the ohm s law equation , i.e. , @xmath10 where @xmath11 is the plasma velocity , @xmath12 is the current density , and @xmath13 is the ohmic resistivity coefficient . in our studies
we are not interested in the acceleration by the electric field resulting from resistivity effects , therefore we have neglected the last term in equation ( [ eq : efield ] ) in the trajectory integration .
the charged particles then feel an electric field @xmath14 besides the magnetic field . substituting the ohm s law , the equation of motion
can be rewritten as @xmath15 .
\label{eq : trajectory}\ ] ] the particle equation of motion ( eq . [ eq : trajectory ] ) was integrated using the 6@xmath16 order implicit runge - kutta - gauss ( rkg ) method ( see * ? ? ?
* e.g. ) with a fixed time step @xmath17 .
the rkg methods are known to conserve the particle energy and momentum in very long integrations @xcite in contrast to the standard 4@xmath16 order runge - kutta method with the adaptive time step based on the 5@xmath16 order error estimator ( see * ? ? ?
* e.g. ) , which is very commonly used .
the interpolation of the local values of the plasma velocity @xmath11 and magnetic field @xmath7 at each step of the integration has been done using cubic interpolation @xcite with our own discontinuity detector based on a total variation diminishing ( tvd ) limiter .
we performed also the integration using linear interpolation which gave us essentially the same statistical results , but with much faster integration times . in the current study
we do not include particle energy losses , therefore test particles can gain or lose energy only through the interactions with the moving magnetized plasma and its fluctuations .
the inclusion of radiative and non - radiative losses and also the back reaction of the particles on the plasma is planned for future studies . for simplicity
, we assume the speed of light to be 20 times the alfvn speed @xmath18 , which defines our plasma in a non - relativistic regime .
the mean density is assumed to be 1 atomic mass unit per cubic centimeter which is compatible with the diffuse interstellar medium ( ism ) density .
the results are presented in units normalized by the assumed light speed and time unit ( which is 1 hour in our simulations ) .
figure [ fig : locations ] presents an evolved 2d configuration of the magnetic field structure with magnetic islands .
we clearly see the merging of islands in some locations and the resulting deformations and/or contractions which provide appropriate conditions for particle acceleration in a similar way to the results obtained with the pic code in ( * ? ? ?
* and references therein ) . in general , we distinguish two kinds of accelerating zones where one or both velocity components ( parallel and perpendicular to the magnetic field ) can increase . in figure [
fig : locations ] , in order to demonstrate the zones which are favorable for a particular type of acceleration , we superimposed on top of the gray texture which represents the magnetic topology , a semitransparent color map showing the locations where acceleration increases the parallel ( red ) or the perpendicular ( green ) velocity component . in yellow areas , both types of acceleration can occur depending on the speed and direction of the test particle with respect to the orientation of the magnetic field lines .
we see that the increase of the parallel velocity component is mostly observed within deformed islands and in current sheets ( see the red and yellow zones in figure [ fig : locations ] ) , while the increase of the perpendicular component is observed mostly near and within the islands and between current sheets ( see the green and yellow zones in figure [ fig : locations ] ) .
this complex behavior is related to the degree of island deformation and the particle direction and speed .
if the island contracts , the particle can increase its parallel component , however , when the island increases its deformation the preferable acceleration usually takes place in the perpendicular direction .
interestingly , we can find islands in which the acceleration does not happen at all ( see , e.g. , the islands above and below the region r3 in figure [ fig : locations ] ) .
this is due to the fact that these islands are not undergoing contraction or deformation processes . in figure
[ fig : particle_histogram ] we present the energy distribution of all particles of the domain of figure [ fig : locations ] at a time corresponding to 1 hour after the injection , when the particles are still accelerating .
the red and blue lines show the number of particles which increase their parallel and perpendicular velocity components , respectively , as a function of the kinetic energy .
in addition , we show the initial thermal distribution of particles ( black line ) . .
blue and red histograms show the number of particles accelerating their perpendicular and parallel velocity components , respectively , at one hour after the particles injection in the system .
the black line exhibits the initial thermal distribution of the injected particles .
[ fig : particle_histogram],scaledwidth=50.0% ] in the particle energy distribution corresponding to the perpendicular acceleration component ( blue ) , we distinguish two modes , one in the low energy range which is related to the initially injected thermal distribution with an extended non - thermal tail , and the other corresponding to the high energy range . for the particle energy distribution corresponding to the parallel acceleration component , we see only one distribution mode with a maximum at high energies . in the low energy range with energies smaller than the particle rest mass energy ( around @xmath19 ) , there is a clear dominance of the perpendicular acceleration .
this creates an anisotropy between the two velocity distributions with respect to the magnetic field .
however , once the particles become relativistic and their kinetic energies become comparable to or larger than the particle rest mass energy , the anisotropy disappears and the acceleration in both directions is equally efficient .
figure [ fig : particle_histograms ] presents the energy distributions of accelerated particles in the selected regions r1 to r6 , as shown in figure [ fig : locations ] .
regions r1 to r3 correspond to islands where both parallel and perpendicular accelerations are observed , while regions r4 to r6 correspond to current sheets .
the energy distributions for regions r1 to r3 are shown in the top panels of figure [ fig : particle_histograms ] .
we notice a dominance of the perpendicular acceleration in the low energy range , similarly as seen in figure [ fig : particle_histogram ] . as we move to higher energies we see a trend to form a saturated tail .
the efficiency of the acceleration of the parallel component increases with the particle energy and at the highest energies the number of particles accelerating in the parallel and perpendicular directions is comparable . in the bottom panels of figure [ fig : particle_histograms ] we present the particle energy distributions for regions within current sheets which are labeled as r4 , r5 , and r6 in figure [ fig : locations ] . in these cases , we also observe an anisotropy between the perpendicular and parallel components in the low energy range with a dominance of the perpendicular acceleration . as in the case of the whole domain
, this anisotropy decreases with increasing energy and , beyond values around the particle rest mass energy both component distributions become comparable .
the high energy range tails of these distributions are steeper than those of the top diagrams of the figure which correspond to magnetic islands .
this is especially well seen in the energy distribution of region r6 .
this indicates that the number of accumulated particles in the high energy range is smaller and therefore the acceleration , although still significant , is less efficient in current sheets than within the islands .
this is apparently due to the fact that ( at least in the 2d case ) the islands are able to retain the particles trapped for longer times
. what could be the dominant mechanism(s ) for the acceleration in these zones ?
the acceleration in the contracting / deforming islands is suggestive of first order fermi processes with the particles bouncing back and forth between converging mirrors as described in @xcite and @xcite , while within and between the current sheets the acceleration mechanism is not as clear . in the following sections , we will examine the acceleration mechanism in these different sites in more detail . in figure
[ fig : event ] , we show an example of a test proton which is trapped in a contracting island and accelerates increasing its parallel component .
the left panel of the figure shows the topology of the magnetic field in a region with a contracting island ( r1 in fig .
[ fig : locations ] ) with the trajectory of the proton superimposed .
as long as the proton remains trapped in the island it orbits around the island center . in the middle panel
we plot the evolution of the parallel and perpendicular velocity components ( red and green lines , respectively ) , and the kinetic energy evolution ( blue line ) of the test particle . in the right panel
, we plot the change of the kinetic energy with the particle x coordinate .
while trapped in the island , the particle orbits around the center and increases its energy after each cycle .
its parallel speed increases while the gyro rotation slows down .
this results in an exponential growth of the kinetic energy of the particle ( see middle panel ) . if we take a closer look in the change of the kinetic energy with the position we see that the increase of energy happens only when the particle moves across the right part of the island .
the left panel showing the magnetic field topology indicates that the field lines in this part are contracting due to the interaction with a small island which is merging with the central one . at the same time , the particle gains more energy after each pass in this zone .
we clearly see how easily the results of @xcite can be reproduced using the mhd approximation , confirming that the process of acceleration in the islands is not restricted to the collisionless physics described by pic codes .
mhd codes present an easier way to study the physics of particle acceleration numerically . we know from shock acceleration theory that particles are injected upstream and allowed to convect into the shock , while diffusing in space so as to effect multiple shock crossings , and thereby gain energy through the first order fermi process .
now , besides this mechanism , they may also experience shock drift acceleration , which is attributed to the grad - b drift when the particle encounters an abrupt change in magnetic fields .
the origin of this effect is due to the net work done on a charge by the lorentz force ( eq . [ eq : efield ] ) in a zone of non - uniform large scale magnetic field .
the principal equation governing this is the scalar product of the particle velocity ( or momentum ) and the acceleration by the convective electric field , @xmath20 . in uniform magnetic fields , the energy gain and loss acquired during a gyroperiod
exactly cancel , so in result no net work is done , @xmath21 .
in contrast , when the gyromotion of a charged particle straddles a discontinuity , the sharp field gradient induces an asymmetry in the time spent by the charge in either side of the discontinuity , so that energy gain and loss do not compensate each other . the compressive nature of the field discontinuity biases the net work done to positive increments in shock encounters between upstream excursions , and
it can be shown that @xmath22 , where the @xmath14 drifts lie in the x - direction , in other words , this energy gain scales linearly with displacement along the drift coordinate x ( i.e. , along the discontinuity ) . thus the energy gained by a particle
depends on how far it drifts _ along _ the front ( see , e.g. , * ? ? ?
* ; * ? ? ?
* and references therein ) . similarly , in the current sheet zones ( regions r4 to r6 )
we may be identifying two distinct mechanisms of acceleration , either first order fermi acceleration in contracting / merging islands which are just forming there ( or even the simple particle scattering between the converging flows entering both sides of the current sheet , as described in * ? ? ?
* ) , or a drift acceleration , as described above .
the islands in current sheets are smaller than those of regions r1 , r2 and r3 , what results in smaller acceleration rates as we see in the lower panels of figure [ fig : particle_histograms ] .
in the zones above and below the current sheets it is possible that we see predominantly a drift acceleration driven by non - uniformities of the magnetic field .
generally , this effect is less efficient than the first order fermi process in merging / contracting islands and results in smaller acceleration rates . in order to better understand these distinct acceleration mechanisms ,
we have also explored the details of the acceleration of a test particle near and within a single ( sweet - parker shaped ) current sheet .
figure [ fig : sp_acceleration ] shows the trajectory and energy evolution of this test particle .
we see that before the particle reaches the current sheet discontinuity it is drifted by the plasma inflow and the increasing gradient of b as it approaches the current sheet . when it enters the discontinuity ( the white part of the trajectory in the left panel ) , it bounces back and forth several times and gains energy ( which increases exponentially as shown in the middle panel of figure [ fig : sp_acceleration ] ) due to head - on collisions with the converging flow , on both sides of the magnetic discontinuity ( i.e. , in a first order fermi process , as described in * ?
at the same time it drifts along the magnetic lines which eventually allow it to escape from the acceleration region .
therefore , we see two mechanisms : a drift acceleration ( dominating outside of the current sheet ) and first order fermi acceleration inside the current sheet .
these processes naturally depend on the initial particle gyroradius , since it determines the amount of time the particle remains in the acceleration zone before escaping .
finally , we may also argue that turbulence above and below the current sheets in figure [ fig : locations ] , far from the islands and diffusion regions , favor second order fermi acceleration mechanisms with particles being scattered by approaching and receding magnetic irregularities .
nevertheless , the first order fermi processes occurring within the islands and current sheets dominate the overall particle acceleration in the system .
the results presented in the previous sections were obtained for 2d models without a guide field .
this means that in this case the magnetic lines creating the islands are closed and a charged particle can be trapped indefinitely in such an island .
the presence of a guide field normal to the plane of figure [ fig : locations ] opens the magnetic loops and allows the charged particles to travel freely in the out - of - plane direction .
moreover , the islands evolve much slower in the presence of a strong guide field . protons in 2d models of reconnection with a guide field strength @xmath23=0.0 and 0.1 ( top and middle panels , respectively ) .
in the bottom panel a fully 3d model with initial @xmath23=0.0 is presented .
the colors show how the parallel ( red ) and perpendicular ( blue ) components of the particle velocities increase with time .
the energy is normalized by the rest proton mass energy .
the background magnetized flow with multiple current sheet layers is at time 4.0 in alfvn time units in all models .
[ fig : energy_2d_3d],title="fig:",scaledwidth=50.0% ] protons in 2d models of reconnection with a guide field strength @xmath23=0.0 and 0.1 ( top and middle panels , respectively ) . in the bottom panel a fully 3d model with initial @xmath23=0.0
is presented .
the colors show how the parallel ( red ) and perpendicular ( blue ) components of the particle velocities increase with time .
the energy is normalized by the rest proton mass energy .
the background magnetized flow with multiple current sheet layers is at time 4.0 in alfvn time units in all models .
[ fig : energy_2d_3d],title="fig:",scaledwidth=50.0% ] protons in 2d models of reconnection with a guide field strength @xmath23=0.0 and 0.1 ( top and middle panels , respectively ) . in the bottom panel a fully 3d model with initial @xmath23=0.0
is presented .
the colors show how the parallel ( red ) and perpendicular ( blue ) components of the particle velocities increase with time .
the energy is normalized by the rest proton mass energy .
the background magnetized flow with multiple current sheet layers is at time 4.0 in alfvn time units in all models .
[ fig : energy_2d_3d],title="fig:",scaledwidth=50.0% ] in figure [ fig : energy_2d_3d ] , we present the time evolution of the kinetic energy of the particles which have their parallel and perpendicular ( red and blue points , respectively ) velocity components accelerated for three models of reconnection .
the upper panel shows the energy evolution for a 2d model without the guide field ( as in the models studied in the previous sections ) .
initially , the particles pre - accelerate by increasing their perpendicular velocity component only .
however , later there is an exponential growth of energy mostly due to the acceleration of the parallel component which stops after the energy reaches values of 10@xmath2410@xmath25 @xmath26 ( where @xmath26 is the proton rest mass energy ) . from that level on
, particles accelerate their perpendicular component only with smaller linear rate in a log - log diagram .
the middle panel shows the kinetic energy evolution in a 2d model with a weak guide field @xmath23=0.1 normal to the plane of figure [ fig : locations ] . in this case
, there is also an initial slow acceleration of the perpendicular component followed by the exponential acceleration of the parallel velocity component .
however , due to the presence of a weak guide field , the parallel component accelerates further to higher energies at a similar rate as the perpendicular one .
this implies that the presence of a guide field removes the restriction seen in the 2d model without a guide field and allows the particles to increase their parallel velocity components as they travel along the guide field , in open loops rather than in confined 2d islands .
this result is reassured by the 3d model in the bottom panel of figure [ fig : energy_2d_3d ] , where no guide field is necessary as the mhd domain in fully three - dimensional . in this case
, we clearly see a continuous increase of both components , which suggests that the particle acceleration behavior changes significantly when 3d effects are considered , where open loops replace the closed 2d reconnecting islands . considering the parametrization we have chosen for our models ,
the gyroradius of a proton becomes comparable to the size of the box when its lorentz factor reaches a value of a few times 10@xmath25 .
the largest islands in the system can have sizes of a few tenths of the size of the box .
these rough estimates help us to understand the energy evolution in figure [ fig : energy_2d_3d ] and the transition from an exponential to a much slower ( linear ) growth rate in the energy around 10 hours .
we note that in the diagrams of this figure , the energy is normalized by the rest mass value , so that in fact , it is the lorentz factor that is plotted . in the case with absence of a guide field ( top panel of fig .
[ fig : energy_2d_3d ] ) , the exponential parallel acceleration stops right before the energy value 10@xmath25 is reached . after this , the rate of acceleration significantly decreases .
this occurs because the larmor radius of the particles has become larger than the sizes of biggest islands .
therefore , from this level on the particles can not be confined anymore within the islands and the first order fermi acceleration ceases .
after that , there is a much slower drift acceleration ( of the perpendicular component only ) caused by the gradients of the large scale magnetic fields .
if a guide field is inserted in such a system , the picture is very similar , except for one detail .
now , since the particles are able to travel along the guide field , their parallel velocity component also continues to increase after the 10@xmath25 threshold ( see the middle panel of fig . [
fig : energy_2d_3d ] ) . of course
, in the 3d model , the particles follow the same trend ( bottom panel of fig . [
fig : energy_2d_3d ] ) .
figure [ fig : energy_guide_field ] exhibits the dependence of the acceleration rate on the out - of - plane guide field strength .
it shows the particle kinetic energy distribution for three models of reconnection with different strengths of the guide field ( @xmath23=0.0 , 0.5 , and 1.0 for the top , middle , and bottom panels , respectively ) 1 hour after the injection .
the initial injected thermal distribution is represented by the blue line .
the particles trajectory integration was performed at the same mhd snapshot , at time 4.0 ( in units of alfvn time ) , for all particles in these models .
we note that at the final state the particle distributions have developed high energy tails which depart from the initial thermal distribution .
however , both the number of accelerated particles to higher energies and the maximum energy at the final time interval strongly depends on the strength of the guide field .
in this paper we have investigated the process of particle acceleration in reconnection zones considering test particles inserted in an isothermal magnetohydrodynamic domain containing a series of reconnection layers , without including kinetic effects .
there were a few goals that motivated our approach .
the first goal was to study if the contraction of magnetic islands , which develop in 2d reconnection layers with finite resistivity , can also occur when the gas pressure is isotropic , as is the case in the mhd regime . in earlier studies , drake and collaborators
@xcite invoked kinetic effects such as the firehose and mirror instabilities to control the development and contraction of the islands in current sheets . in the present work , after obtaining the formation of contracting / deformed islands in current sheets in a nearly incompressible mhd domain , the second goal was to investigate the acceleration of test particles particularly in the direction parallel to the large scale magnetic fields during island contractions .
we were able to identify several regions where such acceleration occurs both in a 2d environment with and without a guide field and in a 3d environment .
also , we were able to identify the nature of the acceleration mechanisms in the different acceleration regions , not only within the contracting islands , but also near and inside the current sheets .
we found that first order fermi processes occur in the islands and also within the current sheets , while drift acceleration due to magnetic field gradients seems to be dominating outside the current sheets .
= 0.0 , 0.5 , and 1.0 ( top , middle and bottom panels , respectively . ) at a time corresponding to 1 hour after the injection .
the initial particle thermal distribution is given by the blue line .
[ fig : energy_guide_field],title="fig:",scaledwidth=50.0% ] = 0.0 , 0.5 , and 1.0 ( top , middle and bottom panels , respectively . ) at a time corresponding to 1 hour after the injection .
the initial particle thermal distribution is given by the blue line .
[ fig : energy_guide_field],title="fig:",scaledwidth=50.0% ] = 0.0 , 0.5 , and 1.0 ( top , middle and bottom panels , respectively . ) at a time corresponding to 1 hour after the injection .
the initial particle thermal distribution is given by the blue line .
[ fig : energy_guide_field],title="fig:",scaledwidth=50.0% ] our results for particle acceleration in 2d mhd models were quite similar to those obtained by @xcite , i.e. , during an island contraction , in our case resulting from the merge with other islands , a particle trapped in it can accelerate and increase its energy exponentially and its parallel velocity component grows while the perpendicular one undergoes a net decrease . during this process
the energy gain is proportional to the particle energy .
these are all necessary conditions in a fermi like process and a close examination of the motion of a test particle in a contracting island shows clear evidence of the first order fermi acceleration process . similarly , exponential acceleration of the parallel velocity component was also detected in regions within the current sheets . a detailed examination of a test particle trajectory within and near a current sheet revealed that outside the discontinuity the particle experiences drift acceleration due to magnetic field gradients and within the current sheet it goes back and forth between the converging flows on both sides of the discontinuity also gaining energy exponentially , just as described in @xcite , while drifting along the magnetic field direction .
@xcite have claimed that particle acceleration in an x - point reconnection in a 3d environment , even in the presence of a guide field , should behave like in 2d systems .
however , we have shown that in the presence of an out - of - plane guide field , this picture changes completely ( see also * ? ? ?
while in 2d mhd models without a guide field the parallel acceleration stops after reaching a certain energy threshold , in 2d mhd models with a guide field this constraint is removed , and particles can continue increasing their parallel speed as they travel along the guide field
. furthermore , in fully 3d mhd models with no guide field the acceleration of the particles exhibits the same trend as in 2d models with a guide field , i.e. , there is no constraint on the acceleration of the parallel speed .
this result is very important as it implies that the overall picture of particle acceleration in 3d reconnection can be very distinct from that in 2d reconnection .
also , it can offer some important clues for particle acceleration in more complex domains , such as in the presence of 3d turbulence where stochastic reconnection is involved ( see * ? ? ? * in prep . ) .
another important result of the present work was to demonstrate that the investigated acceleration mechanisms in reconnection sites ( both drift and fermi processes ) can be present in any environment whose evolution can be approximated by a nearly ideal mhd description as employed here , therefore , without the necessity of invoking kinetic instabilities or anomalous resistivity effects to control the pressure anisotropy or the reconnection rate .
the nearly non - resistive mhd approach with test particle injection offers a possibility of exploring more realistic systems . even though pic codes in general allow the study of plasma processes in greater detail , they present some disadvantages with regard to an mhd description , such as the necessity of a much higher complexity in the 3d modeling , for instance , of turbulence . a final remark is in order .
@xcite investigated particle acceleration in reconnection zones and concluded that mhd should not be a good approximation to describe the whole process of acceleration .
however , their 3d numerical simulations were performed in a fully resistive mhd regime .
therefore , they obtained very efficient particle acceleration due to the high electric field induced by resistivity and an absorption of most of the available magnetic energy by the electrons in a very small fraction of the characteristic time of the mhd simulation .
this led them to conclude that resistive mhd codes are unable to represent the full extent of particle acceleration in 3d reconnection . in the present work
, we have investigated particle acceleration in a nearly ideal mhd regime where only numerical resistivity is present . in this case , the contribution of a resistivity induced electric field is negligible when compared to the advection component , namely , the electric field resulting from the plasma motion in the magnetized medium , @xmath27 .
moreover , since there is no important dissipation effects , our model is scale independent and allows for dimensionless units , so that it is straightforward to rescale the configuration to reproduce different astrophysical environments .
the main drawback of the numerical resistivity is the creation of a hole in the center of the magnetic islands , since the numerical modeling can not handle properly the strongly curved magnetic lines over a few cells . as a consequence ,
numerical dissipation removes part of the magnetic flux in this region .
the importance of the present study of acceleration in the mhd regime is motivated by the fact that magnetic reconnection in 3d becomes fast according to the model in @xcite ( see also * ? ? ?
this model has been successfully tested via numerical simulations in @xcite which confirmed that thick reconnection layers form , where magnetic energy is transferred into energy of contracting loops . in the present paper , we have confirmed that these loops ( or magnetic irregularities ) can act as the places of efficient particle acceleration providing the support for the mechanism first discussed in @xcite . at the same time
, we see that the process of acceleration does not amount to only the process of first order fermi acceleration outlined in the aforementioned paper .
our simulations show a complex interplay of different acceleration processes , which motivates for further studies of the acceleration in the presence of realistic turbulent reconnection ( see * ? ? ? * ) .
the present study has clearly an exploratory character .
it testifies that the acceleration in reconnection regions may be more complex than it may be inferred from earlier simplified 2d studies .
further study of the acceleration in the mhd regime is absolutely essential , as ubiquitous astrophysical turbulence is expected to induce magnetic reconnection all over astrophysical volumes and should be explored in more detail . in this situation
the acceleration of particles by reconnection may play a vital role , the extend of importance of which can be evaluated from further research @xcite .
the results of the paper can be very briefly summarized as follows : * advances in the understanding of magnetic reconnection in the mhd regime , in particular , turbulent magnetic reconnection in @xcite model motivate the studies of whether the reconnection in this regime can accelerate energetic particles . * contracting magnetic loops in magnetic reconnection in 2d , in the mhd regime , provides the acceleration which successfully reproduces the results obtained earlier with more complicated pic codes , which proves that the acceleration in reconnection regions is a universal process which is not determined by the details of plasma physics .
* acceleration of energetic particles in 2d and 3d shows substantial differences , which call for focusing on realistic 3d geometries of reconnection .
our study also shows that apart from the first order fermi acceleration , additional acceleration processes interfere .
gk and emgdp acknowledge the support by the fapesp grants no .
2006/50654 - 3 and 2009/50053 - 8 , and the cnpq grant no .
300083/94 - 7 , and al thanks the nsf grant ast 0808118 , nasa grant nnx09ah78 g and the support of the center for magnetic self organization .
this research was also supported by the project tg - ast080005n through teragrid resources provided by texas advanced computing center ( tacc : http://www.tacc.utexas.edu ) .
part of the computations presented here were performed on the galera supercomputer in the academic computer centre in gdask ( task : http://www.task.gda.pl/ ) .
baring , m. g. & summerlin , e. j. 2009 , in aip conf .
1183 , shock waves in space and astrophysical environments : 18th annual international astrophysics conference , ed .
x. ao & g. z. r. burrows ( melville , ny : aip ) , 74 ciaravella , a. , & raymond , j. c. 2008 , , 686 , 1372 de gouveia dal pino , e. m. & lazarian , a. 2000 , , 536 , l31 de gouveia dal pino , e. m. & lazarian , a. 2001 , , 560 , 358 de gouveia dal pino , e. m. & lazarian , a. 2005 , , 441 , 845 de gouveia dal pino , e. m. , piovezan , p. p. & kadowaki , l. h. s. 2010a , , 518 , a5 de gouveia dal pino , e. m. , kowal , g. , kadowaki , l. h. s. , piovezan , p. , & lazarian , a. 2010b , international journal of modern physics d , 19 , 729 drake , j. f. , swisdak , m. , schoeffler , k. m. , rogers , b. n. , & kobayashi , s. 2006 , , 33 , 13105 drake , j. f. , opher , m. , swisdak , m , & chamoun , j. n. 2010 , , 709 , 963 kowal , g. , lazarian , a. , vishniac , e. t. , & otmianowska - mazur , k. 2009 , , 700 , 63 kowal , g. , de gouveia dal pino , e. m. & lazarian , a. 2011 , in prep .
la rosa , t. n. , shore , s. n. , joseph , t. , lazio , w. , & kassim , n. e. 2006 , journal of physics conference series , 54 , 10 lazarian , a. 2005 , in aip conf .
proc . ser .
784 , magnetic fields in the universe : from laboratory and stars to primordial structures , ed .
e. m. de gouveia dal pino , g. lugones , & a. lazarian ( melville , ny : aip ) , 42 lazarian , a. 2006 , astronomische nachrichten , 327 , 609 lazarian , a. & vishniac , e. t. 1999 , , 517 , 700 lazarian , a. , petrosian , v. , yan , h. & cho , j. 2003 , review at the nbsi workshop `` beaming and jets in gamma ray bursts '' , copenhagen , august 12 - 30 , 2002 ( arxiv : astro - ph/0301181 ) lazarian , a. , vishniac , e. t. , & cho , j. 2004 , , 603 , 180 lazarian , a. & opher , m. 2009 , , 703 , 8 lazarian , a. & desiati , p. 2010
, , 722 , 188 lekien , f. , & marsden , j. 2005 , international journal for numerical methods in engineerings , 63 , 455 litvinenko , y. e. 2003 , , 216 , 189 lugones , g. 2011 , procs . of science , in press mackay , f. , marchand , r. , & kabin , k. 2006 , , 111 , a106205 melrose , d. b. 2009 , in press ( arxiv:0902.1803 ) onofri , m. , isliker , h. , & vlahos , l. 2006 , , 96 , 151102 parker , e. n. 1957 , , 62 , 509 press , w. h. , teukolsky , s. a. , vetterling , w. t. , & flannery , b. p. 1992
, _ numerical recipes in c ( 2nd ed . ) : the art of scientific computing _
, cambridge university press , new york , ny , usa sanz - serna , j. m. & calvo , m. p. 1994 , _ numerical hamiltonian problems _ , chapman & hall , london , uk sweet , p. a. 1958 , in iau symp . 6 , electromagnetic phenomena in cosmical physics , ed .
b. lehnert ( cambridge : cambridge univ .
press ) , 123 yugo , h. & iyemori , t. 2001 , , 106 , 26075 zhang , b. & yan , h. 2011 , , 726 , 90
|
the magnetic fields can change their topology through a process known as magnetic reconnection .
this process in not only important for understanding the origin and evolution of the large - scale magnetic field , but is seen as a possibly efficient particle accelerator producing cosmic rays mainly through the first order fermi process . in this work we study the properties of particle acceleration in reconnection zones and show that the velocity component parallel to the magnetic field of test particles inserted in nearly non - resistive magnetohydrodynamic ( mhd ) domains of reconnection without including kinetic effects , such as pressure anisotropy , the hall term , or anomalous effects , increases exponentially . also , the acceleration of the perpendicular component is always possible in such models .
we have found that within contracting magnetic islands or current sheets the particles accelerate predominantly through the first order fermi process , as previously described , while outside the current sheets and islands the particles experience mostly drift acceleration due to magnetic fields gradients .
considering two dimensional mhd models without a guide field , we find that the parallel acceleration stops at some level .
this saturation effect is however removed in the presence of an out - of - plane guide field or in three dimensional models .
therefore , we stress the importance of the guide field and fully three dimensional studies for a complete understanding of the process of particle acceleration in astrophysical reconnection environments .
|
nuclei with the equal numbers of neutrons and protons ( @xmath0 ) have become a special ground to investigate neutron - proton ( _ np _ ) correlations . in @xmath0 nuclei ,
neutrons and protons occupy the same shell - model orbits , leading to large spatial overlaps between neutron and proton wave functions .
the _ np _ pairing is further enhanced in odd - odd @xmath0 nuclei due to odd neutron and odd proton correlation .
the _ np _ residual pairing interaction can be equally important as the neutron - neutron ( _ nn _ , @xmath6 ) and proton - proton ( _ pp _ , @xmath7 ) pairings .
while _ nn _ and _ pp _ pairings have been well established in terms of the simple bcs model , the _ np _ pairing is still an open question that has motivated many recent theoretical and experimental works @xcite .
in contrast to _ pp _ and _ nn _ correlations , the _ np _ pairing can form isovector ( @xmath8 ) and isoscalar ( @xmath9 ) configurations .
the isovector channel manifest itself in a similar fashion to like - nucleon correlations , which can be evidenced with the presence of nearly degenerate @xmath2 isobaric triplets along the @xmath0 line @xcite .
the existence of the @xmath1 isoscalar component has also been shown theoretically , e.g. , in ref .
@xcite . in light @xmath0 nuclei
, ground states usually favor the lowest @xmath1 isospin quantum number , while @xmath2 states appear at very high excitation energies . with increasing mass number , however ,
@xmath1 and @xmath2 bands begin to coexist at low excitation energies @xcite . in particular , the isospin structures of medium - mass _ fp _ shell nuclei are very interesting , where @xmath2 states can be lower than @xmath1 states .
@xmath2 ground states have been expected systematically @xcite .
recent advances in experimental techniques enable us to study the structures of the odd - odd @xmath0 nuclei in the _ fp _ shell at high spins and high excitation energies .
this provides an unique opportunity to investigate the interplay between @xmath1 and @xmath2 bands .
moreover , the high - precision measurements of level schemes provide useful information for discussion of the isospin asymmetry in their excited states at different angular momenta @xcite .
experimentally , isospin effects can be discussed by defining the mirror energy differences ( med ) and triplet energy differences ( ted ) between analogous states in mirror pairs and @xmath2 triplets , respectively @xcite .
the differences have their origins from the coulomb interaction and charge dependence in strong force , while the former usually plays a dominant role . in the _ fp _ shell , the coulomb shifts of single - particle orbits are relatively small @xcite .
this allows us to extract the effect from the charge symmetry breaking in the strong force for the many - body systems of nuclei .
the purpose of this paper is to study the structures and properties of odd - odd @xmath0 nuclei in the _ fp _ model space by the shell - model diagonalization method .
the many - body interaction used in the effective hamiltonian has been derived from a high - precision charge - dependent bonn ( cd - bonn ) nucleon - nucleon ( _ nn _ ) potential @xcite using the so - called folded - diagram method @xcite .
the obtained interaction retains the charge - independence breaking effects in the _ nn _ potential . the contribution from the coulomb field is also included .
hence the effective hamiltonian does not conserve isospin . in section
ii , we describe briefly the folded - diagram method and the construction of the residual interaction . in section iii , the level structures of the odd - odd @xmath0 nuclei in the @xmath10 shell are calculated and discussed . in section iv , we discuss the electromagnetic properties of the odd - odd @xmath0 nuclei with analyzing the isospin structures of the nuclei .
we start from the mean - field approximation with the perturbation expansion . in this approximation , it is convenient to reduce the degree of freedom of the hilbert space for the many - fermion system of nuclei . for large - scale shell - model diagonalizations ,
the determination of the model space is a crucial step .
the model space can be defined with the projection operator , @xmath11 where @xmath12 is the dimension of the model space , and wave functions @xmath13 are eigenfunctions of the unperturbed hamiltonian , @xmath14 , with @xmath15 being the kinetic energy and @xmath16 an appropriate one - body potential .
the remaining space ( so - called excluded space ) can be expressed by @xmath17 the system of the full hamiltonian with eigen energies @xmath18 and wave functions @xmath19 that are determined by the full schrdinger equation of @xmath20 can be reduced within the model space by projection @xcite , @xmath21 with @xmath22 and @xmath23 is the energy of the assumed core left behind in practical calculations .
@xmath24 is the mode - space - dependent effective hamiltonian . in the shell model context
, the effective hamiltonian can be decomposed into two parts , @xmath25 where @xmath26 is the effective one - body hamiltonian with @xmath27 being single - particle energies ( spe ) .
the @xmath27 values can be obtained from the separation energies of valence particles .
usually , the effective interaction @xmath28 takes the form of pure two - body interactions .
it can be expressed as two body matrix elements in harmonic oscillator ( ho ) basis .
the shell model provides the microscopic foundation to study the excitations of nuclei .
the cohen - kurath @xcite and universal sd ( usd ) @xcite interactions have been proved to be quite successful and useful for the @xmath29- and @xmath30- shell calculations , respectively . in past decades
, extensive theoretical works have been conducted to improve the calculations of the shell model for @xmath3 nuclei in which exact diagonalizations have become possible due to the advance in computing techniques .
the central spirit of shell model calculations is to find a unified interaction for a given mass region .
for the _ fp _ nuclei , various interactions have been proposed , such as the kuo - brown ( kb ) renormalized g matrix @xcite and its modified versions with monopole centroid corrections @xcite , the fpd6 analytic two - body potential @xcite and the gxpf interaction @xcite .
these interactions have been extensively tested with the successful descriptions of rotational collectivities of nuclei in the upper part of the @xmath3 shell ( e.g. , @xmath31cr ) @xcite .
however , these interactions conserve isospin symmetry . in this work
, we present a new isospin - nonconserving effective interaction derived microscopically from the underlying @xmath32 potential .
traditionally , effective residual interaction is derived from phenomenal fit to observations @xcite ( e.g. , the usd interaction for the @xmath30 shell . ) or from the evaluation of the g reaction matrix @xcite .
although the effective interaction can in principle be derived from true @xmath32 interaction , it has to be renormalized to include the contributions from the excluded configurations and to avoid the hard core of the bare @xmath32 potential .
renormalization methods have been proposed for the calculation of effective interaction from underlying meson - exchange potentials , like the momentum - space renormalization group decimation method ( @xmath33 approach ) @xcite and the unitary correlation operator method ( ucom ) @xcite . in @xmath33 approach , high - momentum contributions are integrated out with the introduction of a cutoff , while ucom describes short - range correlations by a unitary transformation . in this paper
, we used the so - called folded - diagram expansion method in which it is more convenient to include effects from excluded configurations .
the folded - diagram expansion is a time - dependent perturbation theory introduced by kuo _
et al . _ for the evaluation of residual interactions
microscopically from bare _ nn _ potentials @xcite .
the short - range repulsion is taken into account by the introduction of g matrix .
the folded ( time - incorrect ) and non - folded ( time - correct ) diagrams are introduced in the evaluation of the time - evolution operator u@xmath34 which can be written using perturbation expansion in the complex - time limit as , @xmath35,\ ] ] where @xmath36 and _ t _ is the time - ordering operator in degenerate or nearly degenerate systems . for a model - space with the dimension @xmath12
, the lowest @xmath12 eigenstates of the true hamiltonian can be constructed using the time - evolution operator with model - space states , @xmath37 , @xmath38 where @xmath39 and @xmath40 is a set of @xmath12 parent states defined as @xmath41 in the folded - diagram decomposition procedure , the wave function @xmath42 can be factorized as @xmath43 where @xmath44 denotes the core and the subscript v means that all interactions in @xmath45 are linked to valence particles .
moreover , diagrams contained in @xmath46 can be expressed in terms of so - called @xmath47-boxes .
the @xmath47-box is made up of non - folded diagrams which are irreducible and valence - linked @xcite .
it is defined as @xmath48 where @xmath49 is the energy of interacting nucleons .
the exclusion operator @xmath50 appears due to the pauli principle preventing interacting nucleons from scattering into states occupied already by other nucleons @xcite . using the generalized time ordering ,
the effective interaction @xmath28 can be expanded in terms of @xmath47-boxes : @xmath51 where @xmath52 denotes a folding sign imposing relative time constraints .
@xmath53 is a set of @xmath47-boxes that have at least two @xmath54 vertices attached to valence lines , satisfying @xmath55 . correspondingly , the solution of eq .
( [ eff ] ) can be derived using an iteration procedure : @xmath56 where the energy dependence of the @xmath47-box can be removed . in practical evaluations ,
non - folded diagrams are calculated firstly and summed to give the @xmath47-box . to treat the strong repulsive core of the bare @xmath32 potential which is unsuitable for perturbation treatments ,
the brueckner @xmath57 reaction matrix is usually used .
the @xmath47-boxes can be expressed in terms of g - matrix vertices .
the @xmath57 matrix is defined by the solution of the bethe - goldstone equation @xcite , @xmath58 satisfying @xmath59 where @xmath60 is the correlated two - body wave function . in the past decade , various high - precision phenomenological meson - exchange potentials have been proposed with reasonable descriptions of the charge symmetry breaking and charge independence breaking @xcite . initiated by weinberg @xcite , the chiral perturbative _ nn _
potential derived from the low energy effective lagrangians of the qcd model shows another promising way .
all the potentials have similar low - momentum behavior . in the present work ,
the bare @xmath32 potential @xmath61 is taken to be a new charge - dependent version of the bonn one - boson - exchange potential as described in ref
for further studies , theoretical works have also been carried out with a n@xmath62lo chiral _ nn _
potential by idaho group @xcite . in the present work ,
we are interested in nuclei in the _ fp _ shell .
the model space contains @xmath63 sub - shells with the core of @xmath64ca .
the @xmath47-boxes are approximated to the third order .
single - particle energies are taken from the experimental excitation energies of the odd neutron in @xmath65ca @xcite . in the cd bonn potential , charge symmetry breaking ( @xmath66 ) and charge independence breaking ( @xmath67 ) effects are embedded in all partial waves with angular momentum @xmath68 . the contribution from the coulomb field
is also added when calculating the g matrix . the derived interaction is composed of three parts : the proton - proton , neutron - neutron and proton - neutron interactions .
no specific mass dependence of the interaction is considered .
we calculated the level structures of odd - odd @xmath0 nuclei in the lower part of _ fp _ shell with the residual interaction derived as above .
the effective hamiltonian is diagonalized with the shell - model code oxbash @xcite .
the hamiltonian matrix is given in proton / neutron formalism , with the neutrons and protons being naturally treated as two kinds of particles .
the calculations of low - lying states in odd - odd @xmath0 nuclei are shown in fig .
[ isomer ] . for nuclei with mass number @xmath69
, calculations have been done in truncated _ fp _ model spaces . while the @xmath70 and @xmath71 states have @xmath2 , other states shown in fig .
[ isomer ] have @xmath1 .
isospin is not an explicit quantum number in our system . for a given state
, the isospin quantum number is assigned through the detailed comparison with neighboring ( @xmath72 ) nuclei .
odd - odd @xmath0 nucleus and neighboring @xmath73 nuclei can form isospin triplet , leading to nearly identical @xmath2 bands .
it can be seen that our calculations reproduce well the relative positions of the @xmath2 and @xmath1 states .
the largest discrepancy compared with experiments is seen for the @xmath74 state in @xmath75sc where the core breaking effect that are not taken into account may be non - negligible . for the next heavier odd - odd @xmath0 nucleus , @xmath76cu
, our calculation can reproduce well the near degenerating @xmath2 and @xmath1 bands , which are not shown in the figure . for heavier odd - odd @xmath0 nuclei ,
e.g. @xmath77ga , contributions from the @xmath78 sub - shell may become significant . for comparison
, we also used an interaction derived from the idaho - a chiral @xmath32 potential @xcite to calculate the low - lying states of the nuclei in fig .
[ isomer ] .
similar results are obtained except @xmath75sc in which the calculated @xmath79 state ( @xmath1 ) is lower than the @xmath70 state by about 47 kev in the idaho - a potential . as mentioned before , for nuclei in @xmath3 shell and above
, @xmath2 ground states exist systematically .
this situation has been studied by satua and wyss @xcite with an extended mean - field model in which the @xmath2 configuration is generated by isospin cranking .
based on the analysis of symmetry energy coefficients , the possible existence of isospin inversion has been extended to @xmath80 by jnecke and odonnell @xcite . in the shell - model hamiltonian ,
different components of two - body interactions are mixed up .
it is not easy to identify the role played by each component .
one possible way is to separate contributions from so - called isovector and isoscalar pairing terms of the shell - model effective interaction @xcite .
such an effort has already been tried @xcite . in ref .
@xcite , authors argued that in the @xmath3 shell the competition between the isovector and isoscalar terms can lower the centroid of the @xmath2 state , leading to a @xmath2 ground state .
the shell model could provide an efficient and qualitative way to search the interplay between various pairing correlations on the isospin structures of self - conjugate nuclei .
the detailed evaluations of these terms would be done in future .
tentative calculations have also been done for light nuclei in the @xmath29- and @xmath30- shell .
results for odd - odd @xmath0 nuclei in these shells are encouraging , showing that the folded - diagram method is valid and efficient for the evaluation of shell - model interaction from _ nn _ potentials .
it has long been a challenging problem that previous realistic shell - model calculations can not give the right @xmath81 ground state in odd - odd @xmath0 nuclei @xmath82b @xcite .
our calculations in the _ psd _ shell with the cd - bonn potential reproduce well the @xmath81 ground state and other low - lying states .
experiments have observed high - spin states in @xmath4v and @xmath5mn up to the @xmath3 band termination @xcite .
our calculations give excellent agreement with experiments .
[ v46 ] shows the calculated level scheme for @xmath4v .
comparisons with experiments @xcite for @xmath1 and @xmath2 yrast bands are given in fig .
the ground - state band has been assigned with the quantum number of @xmath83 and @xmath1 @xcite .
recent experiments have deduced the @xmath2 ground - state bands in @xmath4v and its isospin - triplet partners , @xmath4ti and @xmath4cr , up to @xmath84 @xcite .
our calculations can reproduce the observations for the @xmath2 ground - state bands of these three nuclei . in @xmath4v , a state with the excitation energy of @xmath85 kev was detected and tentatively assigned as @xmath86 by garrett _
the state corresponds to our calculated @xmath87 state with the energy of 8309 kev in which the isospin quantum number was assigned as @xmath2 .
calculated level scheme in @xmath4v .
the comparison with experimental observations is shown in fig .
[ v46a ] . ] for the next heavier @xmath0 nucleus , @xmath5mn , we chose a truncation by allowing a maximum of two particles being excited to higher sub - shells above the @xmath88 orbit .
the truncation is valid since configurations for low - lying states are governed by the @xmath3 sub - shell .
[ scheme ] gives calculated and experimental positive - parity states in @xmath5mn .
the experiment @xcite has identified the @xmath1 band up to the @xmath3 band termination and the @xmath2 band up to @xmath89 .
the observed @xmath90 state at 4874 kev and @xmath91 state at 6460 kev were assigned @xmath1 by oleary _ et al .
our calculations reproduce the @xmath1 and @xmath2 states .
@xmath2 states can be selected from the calculations of electrical quadruple ( e2 ) and magnetic dipole ( m1 ) transition properties that will be discussed in the next section .
comparison between calculated and experimental @xmath2 and @xmath1 yrast bands in @xmath4v .
the experimental data are taken from refs .
] in the framework of the deformed nilsson diagram , low - energy configurations in @xmath5mn have odd proton and odd neutron occupying the @xmath92 $ ] orbit of the @xmath3 sub - shell , leading to low - lying @xmath93 ( @xmath1 or 1 ) and @xmath94 ( @xmath1 ) configurations that have been identified experimentally @xcite .
our calculations have well reproduced the observed states , as shown in fig .
[ scheme ] .
experiments observed that the @xmath94 state ( @xmath1 ) is a @xmath95 m isomer with an excitation energy of 229 kev @xcite .
our calculation gives a energy of 322 kev for the isomer .
its decay is highly hindered due to the @xmath96 selection rule .
low - lying metastable @xmath1 states also exist in other odd - odd @xmath0 nuclei shown in fig .
[ isomer ] , due to large angular momentum gaps between @xmath70 ground states and low - lying high - spin states .
the @xmath93 band ( @xmath1 ) in @xmath5mn has been observed with three members of @xmath97 , @xmath98 and @xmath99 , starting at the excitation energy of 651 kev for the @xmath100 state . in present calculations
, the lowest @xmath100 state appears at 661 kev , forming a band as shown in fig .
[ emb ] . also from the properties of inter - band m1 transition properties (
discussed later ) , it can be obtained that the calculated band built on the @xmath100 state corresponds for the observed @xmath101 band .
the recent experiment @xcite assigned new quantum numbers , @xmath102 and @xmath1 , for the observed state at 1917 kev .
our calculation confirms the @xmath99 state with a calculated energy of 1813 kev . calculated and experimental @xcite lowest @xmath2 and @xmath1 states in @xmath5mn . the @xmath1 states from @xmath103
to @xmath104 forms the @xmath94 band . ]
three negative - parity states with possible @xmath105 and @xmath106 assignments were observed in @xmath5mn @xcite . this band has been tentatively interpreted as an octupole vibration band coupled to the @xmath107 isomeric state by svensson _
_ @xcite . in the shell model ,
the negative - parity band could be generated in the @xmath108 shell with one particle exciting from the @xmath109 orbital .
bands in @xmath5mn .
calculated decay strengths ( in @xmath110 ) are shown along with the transition arrows . [ emb ] ]
the electromagnetic ( em ) transition properties of excited states can provide other detailed structure information .
particularly , em properties play important roles in the identifications of various bands . in the following
, we will investigate e2 and m1 transition properties in @xmath4v and @xmath5mn , with a detailed analyse for @xmath5mn . in practical shell - model calculations ,
transitions are treated in the truncated configuration .
the wave functions ( @xmath111 ) of nuclear states are in fact the projections of the full - hamiltonian wave functions ( @xmath112 ) onto the model space with introducing an effective interaction as described by eq .
( [ proj ] ) .
hence , renormalized em transition operators should be used to include effects from the excluded space , @xmath113 where @xmath114 and @xmath115 indicate the initial and final states .
cccccccc & & & + @xmath116 & ho & ws&skcsb&ho & ws & skcsb&exp + + @xmath117 & 45&42 & 37&42&39&35 & + @xmath118 & 151 & 143&126&142&133&118&@xmath119 + @xmath120 & 159&151&133&150&141&125 & + @xmath121 & 119&114&101&114&107&95 & + @xmath122 & 58 & 55&49&55&52&46 & + + @xmath123 & 214 & 205&178&203&191&167&@xmath124 + @xmath125 & 189&182 & 158&180&170&148&@xmath126 + @xmath127 & 168&163&141&160&151&132 & + @xmath128 & 85&83&71&81&77&67 & + @xmath129 & 4.5&4.4 & 3.9&4.3&4.1&3.7 & + .calculated reduced transition strengths of m1 ( in @xmath130 ) and e2 ( in e@xmath62@xmath131 ) transitions in @xmath4v .
experimental data are taken from refs .
[ cols="<,^,^,^,^,^,^,^ " , ] with the folded - diagram method , effective model - dependent transition operators can de derived microscopically @xcite . however , a simple way to evaluate the renormalization effect is to introduce the effective charges of nucleons @xcite .
for example , the effective operator for e2 transitions can be given as , @xmath132 where @xmath114 sums over all particles and @xmath133 ( with @xmath134 ) are renormalized effective charges for protons and neutrons , respectively .
table [ be2 ] gives the shell - model predictions of b(e2 ) values for @xmath94 and @xmath93 bands ( @xmath1 ) in @xmath5mn .
two different groups of empirical effective charges have been used .
one is the commonly - used effective charges of @xmath135 and @xmath136 .
the other group with @xmath137 and @xmath138 was extracted from the b(e2 ; @xmath139 ) values of @xmath140 mirror nuclei @xmath141fe and @xmath141mn by du rietz _
the wave functions of the shell model are calculated with three different bases , i.e. , the harmonic oscillator ( ho ) , the woods - saxon ( ws ) potential and the skyrme force ( skcsb ) by brown _
@xcite in which a charge - symmetry - breaking term has been added . in the @xmath142 band ( @xmath1 ) of @xmath5mn , one e2 strength has been determined @xcite .
the calculated b(e2 , @xmath143 ) value agrees with the measured lower limit .
as seen from table [ be2 ] , calculated decay strengths are relatively stable for different choices of effective charges and single - particle wave functions . reduced e2 decay strengths for @xmath4v are shown in table [ v46bem ] in which the results calculated with the ho single - particle wave function and effective charges of e@xmath144=1.5e and e@xmath145=0.5e are given .
decay scheme and b(e2 ) strengths ( in e@xmath62@xmath131 ) of the proposed t=1 band in @xmath5mn . calculated and experimental strengths @xcite are shown on the right- and left - hand sides of the transition lines , respectively . ]
[ ema ] shows the calculations of electrical quadruple transitions for the @xmath2 band in @xmath5mn .
results were calculated with @xmath135 and @xmath136 and ho single - particle wave functions .
three side bands identified experimentally are shown in the figure .
two bands with @xmath1 on the right side of the figure have been assigned with @xmath83 and @xmath146 , respectively , by sahu _
the proposed @xmath147 state ( @xmath1 ) at 5027 kev , and the @xmath148 state ( @xmath2 ) at 4386 kev have not been observed .
calculated @xmath149 and @xmath150 states are recognized as the observed @xmath90 state at 4874 kev and the @xmath91 state at 6460 kev , respectively .
both states have the isospin of @xmath1 since m1 transition strengths are rather weak compared with inner - band e2 transitions .
the magnetic dipole ( m1 ) operator is dominated by the isovector spin ( @xmath151 ) term .
strong m1 transitions have been observed only in light nuclei . in the _ fp _ shell , due to the coexistence of @xmath1 and @xmath2 bands at low excitation energies , strong isovector ( @xmath152 ) m1 transitions
have been expected @xcite .
we take the g - factor values of free nucleon and the ho basis to investigate the isoscalar and isovector m1 transition properties in @xmath4v and @xmath5mn .
lccc & b(e2 ) & + @xmath153&cal . & cal . & exp .
+ + @xmath154 & & 2.70 & + @xmath155 & 0.28 & 2.27 & @xmath1566.7 + @xmath157 & 0.78@xmath158 & 2.94 & @xmath159 + @xmath160 & 0.70 & 3.13 & @xmath161 + @xmath162 & 1.46 & 1.98 & @xmath1630.24 + @xmath164 & 0.30 & 0.93 & + @xmath165 & 1.66 & 1.31 & + @xmath166 & 0.93@xmath167 & 0.48 & + @xmath168 & 0.43 & 0.31 & + @xmath169 & 3.76 & 0.04 & + + @xmath170 & 2.0 & 8.65@xmath171 & + @xmath172 & 0.34 & 4.67@xmath167 & + @xmath173 & 7.5 & @xmath174 & + @xmath175 & 0.08 & 0.023 & + @xmath176&35 & 0.06 & + part of calculated m1 transition strengths for @xmath4v and @xmath5mn are shown in table [ v46bem ] and [ bm1 ] , respectively .
available experimental data are given for comparison .
our calculations clearly show that @xmath152 , m1 transitions with @xmath177 are enhanced , while other @xmath152 , m1 transitions are retarded due to the _ k _ quantum number selection ( m1 transitions with @xmath178 should be highly hindered ) . in @xmath4v , four strong isovector m1 transitions have been observed .
they were reproduced well in our calculations . in @xmath5mn ,
the uncertainty in experimental data is very large @xcite
. calculated @xmath177 inter - band m1 transitions ( @xmath152 ) are shown in fig .
the strong m1 transitions have also been interpreted by a simple picture of quasideuteron configurations of odd neutron and odd proton in @xmath5mn @xcite .
another exciting problem involving odd - odd @xmath0 nuclei is the study of super - allowed decays .
the @xmath179 values of ground - state fermi ( @xmath180 ) decays would be the same for analogous transitions if the conserved vector current ( cvc ) hypothesis is valid @xcite .
odd - odd @xmath0 nuclei provide the best ground to test the understanding of the electroweak interaction . as discussed in ref .
@xcite , the isospin - symmetry breaking effect has significant contribution on the asymmetry in analogous transitions . in our hamiltonian ,
the isospin symmetry breaking is treated microscopically , which can provide a basis to quantify the effect .
however , it is beyond the scope of this paper .
a microscopical _ fp_-shell interaction has been constructed from the high - precision cd bonn _
nn _ potential using the folded - diagram renormalization method . in the cd bonn potential ,
both charge symmetry breaking and charge independence breaking are well described . the coulomb field is also included in calculating the effective interaction .
hence , in our effective hamiltonian , isospin symmetry is naturally broken .
the isospin - nonconserving interaction has been used to investigate the level structures of odd - odd self - conjugate nuclei in the lower part of _ fp _ shell .
excellent agreements between calculations and experiments have been obtained .
particularly , the interaction gave a good description of the isospin structures in odd - odd @xmath0 nuclei , reproducing well the relative positions of @xmath1 and @xmath2 bands .
further studies done with the idaho - a chiral perturbation potential showed similar results .
moreover , calculations show that effective interactions derived with the folded - diagram method can reproduce the low - lying structures of odd - odd @xmath0 nuclei in other major shells , including the famous case of @xmath82b in _ p _ shell .
our calculations reproduce the high - spin isomeric states in the _ fp _ shell .
in particular , we have performed detailed shell - model calculations for the positive - parity states in @xmath4v and @xmath5mn .
em transition properties and isospin structures in odd - odd @xmath0 nuclei have been predicted and discussed .
we can construct a one - to - one correspondence between calculated results and experimental data with the comparisons of excitation energies and electromagnetic transition properties .
our study shows that the predictions of this _ ab initio _ interaction is encouraging .
it is very interesting to test its performance in other _
_ shell nuclei .
further studies will be carried out in our future works .
the authors are grateful to prof .
b.a . brown for providing us the code oxbash on the request .
this work has been supported by the natural science foundations of china under grant nos .
10525520 and 10475002 , the key grant project ( grant no .
305001 ) of ministry of education of china .
we also thank the pku computer center where numerical calculations have been done .
dean and m. hjorth - jensen , rev .
phys . * 75 * , 607 ( 2003 ) .
macchiavelli , _ et al .
c * 61 * , 041303(r ) ( 2000 ) .
w. satua and r. wyss , phys .
lett * b393 * , 1 ( 1997 ) ; phys .
lett * 86 * , 4488 ( 2001 ) .
j. jnecke and t.w .
odonnell , phys . lett . *
b605 * , 87 ( 2005 ) , and reference therein .
p. vogel , nucl .
a * 662 * , 148 ( 2000 ) .
i. schneider , _ et al .
c * 61 * , 044312 ( 2000 ) . c. friener , _ et al .
c * 60 * , 011304 ( 1999 ) .
a. schmidt , _ et al .
c * 62 * , 044319 ( 2000 ) .
svensson , _ et al .
c * 58 * , r2621 ( 1998 ) .
oleary , _ et al . , _
, phys . lett . *
b525 * , 49 ( 2002 ) .
oleary , _ et al .
_ , phys . lett . *
b459 * , 73 ( 1999 ) .
n. pietralla _
et al . _ ,
phys . rev .
c * 65 * , 024317 ( 2002 ) .
s. skoda , _ et al .
c * 58 * , r5 ( 1998 ) .
lenzi , _ et al .
c * 60 * , 021303 ( 1999 ) .
a. petrovici , k.w .
schmid , and a. faessler , nucl .
* a647 * , 197 ( 1999 ) . a. poves and g. martnez - pinedo , phys
b430 * , 203 ( 1998 ) .
f. brandolini _
et al . _ ,
c * 64 * , 044307 ( 2001 ) .
von brentano _
_ , nucl . phys . *
a682 * , 48c ( 2001 ) .
s.m . lenzi _
et al . _ ,
* 87 * , 122501 ( 2001 ) .
et al . _ ,
lett . * 87 * , 132502 ( 2001 ) .
zuker , s.m .
lenzi , g. martnez - pinedo , and a. poves , phys .
lett . * 89 * , 142502 ( 2002 ) .
r. machleidt , phys .
c * 63 * , 024001 ( 2001 )
. t.t.s . kuo and e.
osnes , folded - diagram theory of effective interaction in nuclei , atoms and molecules , springer lecture notes in physics , vol .
364 ( springer - verlag , berlin , 1990 ) .
s. cohen and d. kurath , nucl . phys . * 73 * , 1 ( 1965 ) .
wildenthal , prog . part
. nucl . phys .
* 11 * , 5 ( 1984 ) .
t.t.s . kuo and g.e .
brown , nucl . phys .
* a114 * , 241 ( 1968 ) .
a. poves and a. zuker , phys .
rep . * 70 * , 235 ( 1981 ) .
e. caurier , g. martnez - pinedo , f. nowacki , a. poves , and a.p .
zuker , rev .
phys . * 77 * , 427 ( 2005 ) .
richter , m.g .
van der merwe , r.e .
julies , and b.a .
brown , nucl . phys . *
a704 * , 134 ( 1991 ) .
m. honma , t. otsuka , b.a .
brown , and t. mizusaki , phys .
c * 65 * , 061301(r ) ( 2002 ) .
e. caurier , a.p .
zuker , a. poves , and g. martnez - pinedo , phys .
c * 50 * , 225 ( 1994 ) .
e. caurier _
et al . _ ,
lett . * 75 * , 2466 ( 1995 ) .
g. martnez - pinedo , a.p .
zuker , a. poves , and e. caurier , phys .
c * 55 * , 187 ( 1996 ) .
bogner , t.t.s .
kuo , and a. schwenk , phys .
rep . * 386 * , 1 ( 2003 ) .
h. feldmeier , t. neff , r. roth , and j. schnack , nucl .
phys . a * 632 * , 61 ( 1998 ) .
m. hjorth - jensen , t.t.s .
kuo , and e. osnes , phys .
rep . * 261 * , 125 ( 1995 ) .
d.r . entem and r. machleidt , phys . lett . *
b524 * , 93 ( 2002 ) . _ table of isotopes _
, 8th ed . , edited by r.b . firestone and v.s .
shirley ( wiley , new york , 1996 ) vol .
a. etchegoyen , w.d.m .
rae , n.s .
godwin , w.a .
richter , c.h .
zimmerman , b.a .
brown , w.e .
ormand , and j.s .
winfield , msu - nscl report no . 524 ( 1985 ) .
brown , prog . part .
phys . * 47 * , 517 ( 2001 ) .
t. siiskonen , m. hjorth - jensen , and j. suhonen , phys .
c * 63 * , 055501 ( 2001 ) .
t. engeland , m. hjorth - jensen , and e. osnes , phys .
c * 61 * , 021302 ( 2000 ) .
r. du rietz _
et al . _ ,
* 93 * , 222501 ( 2004 ) .
brown , w.a .
richter , and r. lindsay , phys .
b483 * , 49 ( 2000 ) .
r. sahu and v.k.b kota , phys .
c * 67 * , 054323 ( 2003 ) .
hardy and i.s .
towner , phys .
c * 71 * , 055501 ( 2005 ) .
|
the structures of odd - odd @xmath0 nuclei in the lower _ fp _ shell have been investigated with a new isospin - nonconserving microscopical interaction .
the interaction is derived from a high - precision charge - dependent bonn nucleon - nucleon potential using the folded - diagram renormalization method .
excellent agreements with experimental data have been obtained up to band terminations .
particularly , the relative positions of @xmath1 and @xmath2 bands were well reproduced .
calculations with another interaction obtained from the idaho - a chiral potential showed similar results .
our calculations also give a good description of the existence of high - spin isomeric states in the @xmath3 sub - shell . as examples , the spectroscopies and isospin structures of @xmath4v and @xmath5mn
have been discussed in detail , with the useful predictions of level structures including electromagnetic properties .
results for @xmath5mn were analyzed where experimental observations are still insufficient .
|
our findings offer insight for constructing an `` early warning system '' for financial markets . by providing a simple instrument to identify similarities to previous states during an upcoming crisis
, one can judge the current situation properly and be prepared to react if the crisis materializes .
another indication for a crisis is given when the correlation structure undergoes rapid changes .
using the similarity measure we were able to classify several typical market states between which the market jumps back and forth .
some of these states can easily be identified in the similarity measure .
however , there are several states in which the market only stays for a short period .
thus , these states are sparsely embedded in time . with a clustering analysis
, we were able to identify these states and disclose a detailed dynamics of the market s state .
a possible application of the similarity measure is risk management .
given the similarity measure , the portfolio manager is aware of periods in which the market behaved completely differently and thus can choose not to include them in his calculations
. he can furthermore identify regions in which the market behaved similarly and refer to these regions when estimating the correlation matrix .
our empirical study is a first step towards the identification of states in financial markets which are a prominent example of complex non - stationary systems .
let @xmath22 be the price of a specific stock and @xmath23 the interval on which the return is calculated . for our study , we chose the arithmetic return , defined as @xmath24 for dataset ( i ) , we chose @xmath23 to be 1 day and calculate the stock returns of each day .
for dataset ( ii ) , we chose @xmath23 as 1 hour . furthermore , we obtain this 1-hour return for every minute of a trading day between 10:45am and 2:45pm .
we obtain the daily data of dataset ( i ) from _ finance.yahoo.com_. the intraday data is obtained from the new york stock exchange s taq database .
sudden changes in drift and volatility can result in erroneous correlation estimates . to address this problem
, we employ a local normalization method @xcite . for each return
@xmath25 we subtract the local mean and divide by the local standard deviation , @xmath26 the local average @xmath27 runs over the @xmath28 most recent sampling points .
for daily data , @xmath29 yields nearly normal distributed time series , as recently discussed @xcite .
our clustering analysis is based on a top - down scheme : all the correlation matrices are initially regarded as a single cluster and then divided into two clusters by the procedure based on the k - means algorithm @xcite .
each division step consists of the following process : 1 .
choose two initial cluster centers from all matrices .
label all other matrices by the more similar cluster center in terms of @xmath30 .
1 . recast two new cluster centers to the `` center of mass '' 2 . re - label
all matrices to their most similar cluster center .
3 . repeat this process until there is no change in labeling .
we stop this division process when the average distance from each cluster center to its members becomes smaller than certain threshold . to identify the typical market states presented in the manuscript , we chose the threshold at 0.1465 as it represents the best ratio between the distances between clusters and their intrinsic radius .
one can obtain finer structures by choosing smaller threshold values , ultimately until all the matrices are identified as different components .
the complete results of the clustering analysis are available in the _ supplementary material _
mcm acknowledges financial support from the fulbright program and from studienstiftung des deutschen volkes .
ts acknowledges support from the jsps institutional program for young researcher overseas visits , grant - in - aid for young scientists ( b ) no .
21740284 mext , japan , and the aihara project , the first program from jsps , initiated by cstp .
ths acknowledges support from project 79613 of conacyt , mexico .
he s thanks the nsf for support .
10 j. voit , _ the statistical mechanics of financial markets _ ( springer , heidelberg , 2001 ) . r. n. mantegna , h. e. stanley , _ physica a _ * 239 * , 255 ( 1997 ) . c. borghesi , m. marsili , s. miccich , _ physical review e _ * 76 * , 026104 ( 2007 ) .
t. f. cooley , v. quadrini , _ the american economic review _ * 91 * , 1286 ( 2001 ) .
d. pelletier , _ journal of econometrics _ * 131 * , 445 ( 2006 ) . x. e. xu , p. chen , c. wu , _ journal of banking & finance _ * 30 * , 1535 ( 2006 ) . m. king , s. wadhwani , _ the review of financial studies _ * 3 * , 5 ( 1990 ) .
s. lee , _ journal of property investment & finance _ * 24 * , 434 ( 2006 ) .
j. bouchaud , m. potters , _ theory of financial risks _ ( cambridge university press , cambridge , 2000 ) .
eckmann , s. o. kamphorst , d. ruelle , _ epl ( europhysics letters ) _ * 4 * , 973 ( 1987 ) .
m. c. casdagli , _ physica d : nonlinear phenomena _ * 108 * , 12 ( 1997 ) .
v. plerou , p. gopikrishnan , e. stanley , _ nature _ * 421 * ( 2003 ) .
b. rosenow , p. gopikrishnan , v. plerou , h. e. stanley , _ physica a _ * 324 * , 241 ( 2003 ) .
proceedings of the international econophysics conference .
h. tastan , _ physica a _ * 360 * , 445 ( 2006 ) .
s. drozdz , j. kwapien , f. grmmer , f. ruf , j. speth , _ physica a _ * 299 * , 144 ( 2001 ) .
r. schfer , n. nilsson , t. guhr , _ quantitative finance _ ( 2009 ) .
l. laloux , p. cizeau , j .-
bouchaud , m. potters , _ physical review letters _ * 83 * , 1467 ( 1999 ) . v. plerou , _
_ , _ physical review e _ * 65 * , 066126 ( 2002 ) .
t. guhr , b. klber , _ journal of physics a : mathematical and general _ * 36 * , 3009 ( 2003 ) . j. schfer , k. strimmer , _ statistical applications in genetics and molecular biology _ * 4 * ( 2005 ) . o. ledoit , m. wolf , _ journal of empirical finance _ * 10 * , 603 ( 2003 ) .
t. w. epps , _ journal of the american statistical association _ * 74 * , 291 ( 1979 ) . m. c. mnnix , r. schfer , t. guhr , _ physica a _ * 389 * , 767 ( 2010 ) . m. c. mnnix , r. schfer , t. guhr , _ physica a _ * 389 * , 4828 ( 2010 ) .
r. schfer , t. guhr , _ physica a _ * 389 * , 3856 ( 2010 ) .
cover story .
( 2008 ) .
http://www.standardandpoors.com/indices/gics/en/us .
j. b. macqueen , _ proc . of the fifth berkeley symposium on mathematical statistics and probability _ , l. m. l. cam , j. neyman , eds .
( university of california press , 1967 ) , vol . 1 ,
281297 .
v. faber , _ los alamos science _ * 22 * , 138 ( 1994 ) .
n. k. tanaka , t. awasaki , t. shimada , k. ito , _ current biology _ * 14 * , 449 ( 2004 ) .
a similar result can be archived using a different approach .
the largest eigenvalue @xmath31 of the correlation matrix @xmath32 describes the collective motion of all stocks .
we can also define the similarity measure by the distance of these eigenvalues , @xmath33 figure [ fig : nateigenvaluediff ] illustrates that this leads to an almost identical result .
the advantage of this technique is that the noise in the correlation matrix only contributes to small eigenvalues @xcite .
thus , by only taking into account the largest one , we filter out the noise .
however , this approach also presumes that the corresponding eigenvector does not change .
our results indicate that the largest eigenvalue almost remains constant , but this might not always be the case . especially during financial crises
. a detailed look of the correlation structure of the 2008 - 2009 crisis can be found in fig .
[ img : nat - stable - within - crisis ] .
while during the crisis , an overall high correlation level dominates , the structure stabilizes for a short time of 3 weeks . during this stable period
, the structure is very similar to state 7 , that occurred just just before the crisis .
some of the correlation structures in fig .
[ img : nat - typicalstates ] look quite similar at first sight .
their distinctiveness can be emphasized by calculating the difference to the average correlation level .
this is shown in fig .
[ fig : matdifference ] . for example , state 3 and 4 look very similar in fig .
[ img : nat - typicalstates ] . however , fig .
[ fig : matdifference ] unveils that the correlation within the energy sector ( e ) is completely diverse .
|
the understanding of complex systems has become a central issue because complex systems exist in a wide range of scientific disciplines .
time series are typical experimental results we have about complex systems . in the analysis of such time series ,
stationary situations have been extensively studied and correlations have been found to be a very powerful tool . yet most natural processes are non - stationary . in particular , in times of crisis , accident or trouble , stationarity is lost . as examples
we may think of financial markets , biological systems , reactors ( both chemical and nuclear ) or the weather . in non - stationary situations analysis
becomes very difficult and noise is a severe problem .
following a natural urge to search for order in the system , we endeavor to define states through which systems pass and in which they remain for short times .
success in this respect would allow to get a better understanding of the system and might even lead to methods for controlling the system in more efficient ways .
we here concentrate on financial markets because of the easy access we have to good data , because of our previous experience and last but not least because of the strong non - stationary effects recently seen .
we analyze the s&p 500 stocks in the 19-year period 1992 - 2010 . here , we propose such an above mentioned definition of state for a financial market and use it to identify points of drastic change in the correlation structure .
these points are mapped to occurrences of financial crises .
we find that a wide variety of characteristic correlation structure patterns exist in the observation time window , and that these characteristic correlation structure patterns can be classified into several typical `` market states '' .
using this classification we recognize transitions between different market states . a similarity measure
we develop thus affords means of understanding changes in states and of recognizing developments not previously seen .
the effort to understand the dynamics in financial markets is attracting scientists from many fields @xcite .
statistical dependencies between stocks are of particular interest , because they play a major role in the estimation of financial risk @xcite . since the market itself is subject to continuous change , the statistical dependencies also change in time .
this non - stationary behavior makes an analysis very difficult @xcite .
changes in supply and demand can even lead to a two phase behavior of the market @xcite . here
, we use the correlation matrix to identify and classify the market state .
in particular we ask , how similar is the present market state , compared to previous states ? to calculate this _ similarity _ we measure temporal changes in the statistical dependence between stock returns . for stationary systems described by a ( generally large ) number @xmath0 of time
series , the pearson correlation coefficient is extremely useful .
it is defined as@xmath1 here the @xmath2 and @xmath3 represent the time series of which the averages @xmath4 are taken over a given time horizon t. @xmath5 and @xmath6 are their respective standard deviations . when calculating the correlation coefficients of @xmath0 stocks , we obtain the @xmath7 correlation matrix @xmath8 , which gives an insight into the statistical interdependencies of the time series under study .
it is necessary to consider data over large time horizons @xmath9 so as to obtain reliable statistics .
this leads to a fundamental problem that arises in the case of _ non - stationary _ systems : to extract useful information from empirical data we seek a correlation matrix from very recent data , in order to provide a good description of current correlation structure .
this is because correlations change dynamically due to the non - stationarity of the process , making it very difficult to estimate them precisely @xcite .
however , if the length @xmath9 of the time series is short , the correlation matrices @xmath8 are noisy . on the other hand , to keep the estimation error low , @xmath9 can be increased , but this leads to a correlation matrix that generally does not describe the present state very well .
various noise reduction techniques provide methods to conquer noise @xcite . in several non - stationary systems , it is possible to obtain a large number of correlated data over time .
such systems include , but are not restricted to , financial markets ( which show non - stationary behavior due to crises ) , biological or medical time series ( such as eeg ) , chemical and nuclear reactors ( non - stationary behavior includes , in particular , accidents ) or weather data . in the following ,
we only consider the financial markets , since we have studied extensively some very high quality data of this system , the non - stationary features of which have been quite striking in the last years .
we propose a definition of a state which is appropriate for such systems and suggest a method of analysis which allows for a classification of possible behaviors of the system .
when @xmath10 , which is the case we are interested in , the correlation matrix becomes singular .
however , one can still make significant statistical statements , e.g. , for the average correlation level whose estimation error decreases as @xmath11 . in the following ,
we focus on correlation matrices @xmath12 and @xmath13 at different times @xmath14 and @xmath15 measured over a _ short _ time horizon .
these have therefore a pronounced random element .
we take these objects as the fundamental states of our system .
we now propose , as a central element , to introduce the following concept of distance between two states .
we define @xmath16 to quantify the difference of the correlation structure for two points in time , where @xmath17 denotes the absolute value and @xmath18 denotes the average over all components .
note that in this case , the random component that is unavoidable in the definition of the states of the system is strongly suppressed by the average over @xmath19 numbers . to apply the above general statements to a specific example ,
we analyze two datasets : ( i ) we calculate @xmath20 based on the daily returns of those s&p 500 stocks that remained part of the s&p during the 19-year period 19922010 , and ( ii ) we study the four - year period 20072010 in more detail based on intraday data from the nyse taq database . since the noise increases for very high - frequency data @xcite , we extract one - hour returns for dataset ( ii ) . for one - hour returns , we consider this market microstructure noise as reasonably weak .
however , sudden changes in drift and volatility are present on all time scales . they can result in erroneous correlation estimates . to address this problem ,
we employ a local normalization @xcite of the return time series in dataset ( i ) .
the results of dataset ( i ) are presented in fig .
[ img : nat - marketsim ] . in this figure , each point is calculated on correlation matrices over the previous two months .
this new representation gives a complete overview about structural changes of this financial market of the past 19 years in a single figure .
it allows to compare the similarity of the market states at different times . to make this procedure concrete ,
consider the following example .
pick a point on the diagonal of fig .
[ img : nat - marketsim ] and designate it as `` now '' .
from this point the similarity to previous times can be found on the vertical line above this point , or the horizontal line to the left of this point .
light shading denotes similar market states and dark shading denotes dissimilar states .
we can furthermore identify times of financial crises with dark shaded areas .
this indicates that the correlation structure completely changes during a crisis .
there are also similarities between crises , as between the `` credit crunch '' that induced the 20082009 financial crisis and the `` market meltdown '' , the burst of the dot - com bubble in 2002 .
a further example is the overall rise in correlation level in the beginning of 2007 .
this event can be mapped to drastic events on the shanghai stock exchange @xcite .
+ + using dataset ( ii ) we are able to obtain a more detailed insight into recent market changes , as shown in fig .
[ img : nat - marketsimintraday ] . this area
is represented by the lower right square in fig .
[ img : nat - marketsim ] . using intraday data
we calculate the correlation matrices on shorter time scales .
we choose a time horizon of one week , which , because it provides insight into changes in the correlation structure on a much finer time scale , enables us to identify a short sub - period within the 20082009 crisis ( in the beginning of 2009 ) during which the market temporarily stabilizes before it returns to the crisis state .
while the correlation structure during the crisis displays an overall high correlation level , the correlation structure of the stable period is similar to the period before the crisis , one of the typical states in a calm period , which is identified from daily data in dataset ( i ) .
this phenomenon might be related to the market s reaction to news about the progress in rescuing the american international group ( a.i.g . ) @xcite .
the correlation structure of this stable period can be found in the _ supplementary material_. the evolutionary structure presented in figs .
[ img : nat - marketsim ] and [ img : nat - marketsimintraday ] illustrate that the correlation matrix sometimes maintains its structure for a long time ( bright regions ) , sometimes changes abruptly ( sharp blue stripes ) , and sometimes returns to a structure resembling a structure the market has experienced before ( white stripes ) .
this suggests that the market might move among several typical market states . to extract such typical market states , we perform a clustering analysis in the results of dataset ( i ) . from our clustering analysis
( see _ methods _ and _ supplementary material _ ) , we find that there are `` hidden '' states sparsely embedded in time , in addition to regimes that dominate the market during a continuous period and are easily found by eye . for this analysis ,
we use disjunct two - month time windows ending at the respective dates . because of the window length
, some financial crashes can not be resolved .
our aim is rather to identify the evolution of the market , which is , in some cases , induced by financial crisis .
we can confirm in fig .
[ img : nat - typicalstates ] that the typical states obtained from the clustering analysis indeed correspond to different characteristic correlation structures . to visualize these characteristic structures , we sort them according to their industry branch using the global industry classification standard ( gics ) @xcite .
the industry branches correspond to the blocks on the diagonal . to visualize the characteristic structures of each state , we calculate its average correlation matrix and sort the companies according to their industry branch , as defined by the global industry classification standard ( gics ) .
the resulting matrices , the industry branches correspond to the blocks on the diagonal .
the correlation between two branches are given by the off - diagonal blocks .
the results are illustrated in fig .
[ img : nat - typicalstates ] .
we can confirm that the typical states obtained from the clustering analysis indeed correspond to different characteristic correlation structures .
our analysis also offers insight into market structure dynamics .
figure [ img : nat - stateevolution ] shows the temporal behavior of the market state .
the market sometimes remains for a long time in the same state , and sometimes stays only for a short time .
the typical duration depends upon the state : some states ( e.g. , state 1 and state 2 ) appear in clusters in time while other states appear more sparsely in time ( e.g. , state 4 ) .
there seems to exist a global trend on a long time scale , although the market state is switching back and forth between states . + in fig .
[ img : nat - typicalstates ] , we can see differences between the states in the correlation between branches as well as in the correlation within a branch .
the correlation within the energy , information technology , and utilities branches is very strong in all states .
state 1 shows an overall weak correlation , while states 3 and 4 feature in addition a strong correlation of the finance branch to other branches .
state 2 shows very unusual behavior : in the period of the dot - com bubble , many branches are anti - correlated with one another . in states 5 , 6 and 7 , the overall correlation level rises , although certain branches , such as energy , consumer staples , and utilities , are either strongly or weakly correlated with other branches . the energy branch ( e )
can be either strongly correlated to the rest of the market , weakly correlated , or even anti - correlated .
therefore we study the histogram of the correlation coefficients @xmath21 .
we present the results in fig .
[ img : histogram_cij ] . in the months
leading up to the credit crunch in october 2008 , we observe a bimodal structure in the histogram .
it corresponds to the time period when the energy branch shows a strong anti - correlation with other branches .
the bimodality suggests that a subset of stocks in this case , predominantly the energy stocks decouples from the rest of the market . during the crash
, the histogram shows a very narrow distribution around large values of the correlation coefficients , which corresponds to state 8 in fig .
[ img : nat - typicalstates ] , where the branch structure is lost almost completely in an overall strongly correlated market .
|
the components of @xmath22 associated with ( [ higgsn ] ) are ( @xmath67 ) w. nahm , phys . lett .
b * 93 * ( 1980 ) 42 .
r. jackiw and c. rebbi , phys .
d * 13 * ( 1976 ) 3398 .
a. gonzalez - arroyo and y. a. simonov , nucl .
b * 460 * ( 1996 ) 429 [ hep - th/9506032 ] .
y. kazama , c. n. yang and a. s. goldhaber , phys .
d * 15 * ( 1977 ) 2287 .
t. t. wu and c. n. yang , nucl .
b * 107 * ( 1976 ) 365 .
|
fermion zero modes for abelian bps monopoles are considered . in the spherically symmetric case
the normalisable zero modes are determined for arbitrary monopole charge @xmath0 .
if @xmath1 the zero modes are zero along @xmath2 half - lines emanating from the monopole .
keywords : abelian gauge theory , bps monopoles , weyl equation , fermion zero modes .
fermion zero modes for bps monopoles can be constructed via the same nahm transform used to obtain the monopoles @xcite .
the construction is , however , cumbersome for magnetic charge @xmath3 . in this letter
we obtain zero modes for abelian bps monopoles .
our approach is to directly integrate the weyl equations in three - dimensional space rather than use nahm s method ( which has been adapted to abelian monopoles in @xcite ) .
the abelian bps equations read @xmath4 where @xmath5 is a real higgs field and @xmath6 is a magnetic field derived from a vector potential @xmath7 .
the maxwell equation @xmath8 implies that the higgs field @xmath5 obeys the laplace equation . the higgs field
@xmath9 with @xmath10 and @xmath11 constant , is harmonic away from @xmath0 singularities @xmath12 ( @xmath13 ) .
physically , the system comprises @xmath0 dirac monopoles each with magnetic charge @xmath11 interacting with a higgs field . here
@xmath10 fixes the asymptotic value of the higgs field .
consider the weyl operators @xmath14 where @xmath15 is the electric charge of the fermion and @xmath16 .
these weyl operators assume a real yukawa coupling in minkowski space .
however , identifying @xmath5 as @xmath17 they are also weyl operators for self - dual monopoles defined in euclidean space .
the dirac quantisation condition requires @xmath18 with @xmath19 integer . if @xmath20 and @xmath21 , @xmath22 has @xmath0 normalisable zero modes . in the @xmath23 case
we have @xmath24 taking the origin as the location of the monopole and @xmath25 denote spherical polar coordinates . here
the dirac string lies on the positive @xmath26-axis .
one can verify that @xmath27 is a zero mode of @xmath22 ( the components of @xmath22 in spherical polar coordinates are given in the appendix ) .
@xmath28 has no zero modes .
this result can be obtained by taking the large @xmath29 limit of the jackiw - rebbi zero mode @xcite for the basic @xmath30 bps monopole after performing a gauge transformation which diagonalises the higgs field .
the fermion density @xmath31 is spherically symmetric and has an integrable singularity at the monopole centre .
the general @xmath0 case is more complicated .
however , in the spherically symmetric case where the positions of the @xmath0 monopoles coincide the higgs field is @xmath32 and @xmath7 is the @xmath23 potential multiplied by @xmath0 . here
@xmath22 has @xmath0 normalisable zero modes : @xmath33 these solutions resemble ( in particular with respect to the @xmath34 and @xmath35 dependence ) known solutions of the dirac equation in the background of abelian monopoles @xcite .
however , our solutions are written directly in terms of trigonometric functions rather than spherical harmonics is even the zero modes can be expressed in terms of standard spherical harmonics . if @xmath0 is odd spin - weighted or monopole harmonics @xcite are required . ] .
our solutions are normalisable with @xmath36 norm @xmath37 . as the zero modes @xmath38 are annihilated by the hamiltonian it is not clear to us whether the presence of the higgs field cures the self - adjointness problem associated with monopole hamiltonians . to address this issue one needs to study the scattering states @xcite .
note that the densities @xmath39 are not spherically symmetric for @xmath1 .
for @xmath40 the first zero mode is zero along the positive @xmath26-axis while the second mode is zero on the negative @xmath26-axis . by taking a suitable linear combination of @xmath41 and
@xmath42 one can obtain a zero mode with a zero along any half - line emanating from the monopole ; the zero mode is axially symmetric about the axis on which the half - line lies .
the @xmath1 zero modes are zero along @xmath2 half - lines .
our @xmath41 and @xmath43 have zeros of strength @xmath2 on the positive and negative @xmath26-axes , respectively ; the remaining @xmath44 modes have lower strength zeros on both the positive and negative @xmath26 axes .
again one can adjust the directions of the @xmath2 half lines by taking different linear combinations of the @xmath0 zero modes .
for a discussion of zeros of fermion zero modes in a different context see @xcite . in general , the @xmath3 zero modes are not axially symmetric even though the @xmath38 are all axially symmetric about the @xmath26-axis . if @xmath45 is a linear combination of the @xmath38 , the density @xmath46 has the form @xmath47 where @xmath48 is a function of @xmath34 and @xmath35 . for @xmath3
one can see that the zero modes vanish at the position of the monopole and the zero modes peak somewhere on the sphere @xmath49 .
the function @xmath50 has up to @xmath2 zeros ; if there are fewer than @xmath2 zeros these are repeated zeros associated with coincident half - lines .
examination of @xmath50 for several zero modes indicates that @xmath50 has a single peak .
for example , the @xmath51 zero mode @xmath52 yields an @xmath50 function with two zeros and a maximum on the equator @xmath53 ( these three points are equally spaced on the equator ) .
the maximum of @xmath50 is a point on the unit sphere except for some axially symmetric solutions where @xmath50 peaks on a circle . in figure 1
spherical plots of @xmath50 are given for an @xmath40 and an @xmath51 zero mode .
for an @xmath40 and @xmath51 zero mode ( the scale depends on the normalisation and hence @xmath10 ) .
the left plot shows @xmath50 for an @xmath40 zero mode ( @xmath54 ) which has one zero and is axially symmetric .
two zeros are visible for the @xmath51 zero mode ( @xmath55 ) on the right .
both zero modes have a single peak . ]
van baal @xcite has given an ansatz which provides a solution of the weyl equation for any solution of the abelian bps equation : @xmath56 satisfies @xmath57 where @xmath58 and @xmath59 satisfies the laplace equation .
this works as the bps equation implies @xmath60 is a scalar and ( [ vanbaal2 ] ) gives @xmath61 .
taking @xmath62 yields our @xmath5 and @xmath7 .
however , @xmath63 is not normalisable ) does provide one normalisable zero mode for a different class of higgs fields ; here @xmath5 has @xmath64 singularities representing @xmath0 positively charged and @xmath0 negatively charged monopoles @xcite .
] though for @xmath65 it agrees with our zero mode @xmath41 .
indeed , @xmath41 is the @xmath65 van baal solution multiplied by @xmath66 . as the van baal construction does not rely on spherical symmetry
this approach may provide information about the @xmath65 limit of the general case where the @xmath0 monopoles are separated .
we have considered fermion zero modes for bps monopoles and have obtained solutions for arbitrary magnetic charge @xmath0 .
the spherically symmetric abelian case we have solved may provide a model for non - abelian magnetic bags ; although higher charge @xmath30 bps monopoles are never spherically symmetric , solutions with approximate spherical symmetry may emerge for large @xmath0 @xcite
. it would be interesting to investigate the extent to which our zero modes approximate the fermion zero modes of magnetic bags .
|
the theory of @xmath0-hypergeometric systems introduced by gelfand - kapranov - zelevinsky @xcite is a vast generalization of that of classical hypergeometric differential equations . as in the case of hypergeometric equations , the holomorphic solutions to
their @xmath0-hypergeometric systems ( i.e. the @xmath0-hypergeometric functions ) admit @xmath1-series expansions ( @xcite ) and integral representations ( @xcite ) .
moreover this theory has deep connections with many other fields of mathematics , such as toric varieties , projective duality , period integrals , mirror symmetry , enumerative algebraic geometry and combinatorics . also from the viewpoint of the @xmath2-module theory ( see @xcite and @xcite etc . ) , @xmath0-hypergeometric @xmath2-modules are very elegantly constructed in @xcite .
for the recent development of this subject see @xcite and @xcite etc . in @xcite , @xcite , @xcite and @xcite etc .
the monodromies of their @xmath0-hypergeometric functions were studied . in
@xcite adolphson generalized the hypergeometric systems of gelfand - kapranov - zelevinsky @xcite to the confluent ( i.e. irregular ) case and proved many important results .
however the construction of the confluent @xmath0-hypergeometric @xmath2-modules is not functorial as in @xcite and @xcite .
this leads us to some difficulties in obtaining the integral representations of their holomorphic solutions ( i.e. the confluent @xmath0-hypergeometric functions ) .
indeed , in the confluent case almost nothing is known about their global properties . in this paper , we first construct adolphson s confluent @xmath0-hypergeometric @xmath2-modules functorially as in @xcite .
note that recently the same problem was solved more completely in saito @xcite and schulze - walther @xcite , @xcite by using commutative algebras .
our approach is based on sheaf - theoretical methods and totally different from theirs
. moreover we also construct an integral representation of the confluent @xmath0-hypergeometric functions by using the theory of rapid decay homology groups introduced recently in hien @xcite and @xcite .
recall that @xmath3 is a finite subset of a lattice @xmath4 and adolphson s confluent @xmath0-hypergeometric system is defined on @xmath5
. then our integral representation of its holomorphic solutions @xmath6 coincides with the one in adolphson ( * ? ? ?
* equation ( 2.6 ) ) , where @xmath7 is a family of real @xmath8-dimensional topological cycles @xmath9 in the algebraic torus @xmath10 on which the function @xmath11 decays rapidly at infinity .
more precisely @xmath9 is an element of hien s rapid decay homology group .
see sections [ sec:3 ] and [ sec:4 ] for the details .
adolphson used the formula without giving any geometric condition on the cycles @xmath9 nor proving the convergence of the integrals . in our theorem [ main ]
we could give a rigorous justification to adolphson s formula ( * ? ? ?
* equation ( 2.6 ) ) by using rapid decay homology cycles .
this integral representation can be considered as a natural generalization of those for the classical bessel and airy functions etc .
note that in the case of hypergeometric functions associated to hyperplane arrangements the same problem was precisely studied by kimura - haraoka - takano @xcite etc .
we hope that our geometric construction would be useful in the explicit study of adolphson s confluent @xmath0-hypergeometric functions . in the proof of theorem
[ main ] , we shall use the method of toric compactifications introduced in @xcite and @xcite for the study of geometric monodromies of polynomial maps .
moreover we introduce proposition [ rth ] which enables us to calculate hien s rapid decay homologies by usual relative twisted homologies . by proposition [ rth ] and lemmas [ ec ] and [ ecl ]
we can calculate the rapid decay homologies very explicitly in many cases .
let @xmath12 be the convex hull of @xmath13 in @xmath14 and @xmath15 the laurent polynomial on @xmath16 defined by @xmath17 . then in section [ sec:5 ] , assuming the condition @xmath18 and using the twisted morse theory we construct also a natural basis of the rapid decay homology group indexed by the critical points of @xmath19 .
furthermore we apply it to obtain a precise formula for the asymptotic expansions at infinity of adolphson s confluent @xmath0-hypergeometric functions .
the formula that we obtain in theorem [ ase ] will be very similar to that of the classical bessel functions . finally in sections [ sec:6 ] and [ sec:7 ] , removing the condition @xmath18 we construct another natural basis of the rapid decay homology group .
we thus partially solve the famous problem in gelfand - kapranov - zelevinsky @xcite of constructing a basis of the twisted homology group in their integral representation of non - confluent @xmath0-hypergeometric functions , in the more general case of confluent ones .
* acknowledgement : * we thank professors k. ando , y. haraoka , t. kohno , a. pajitnov and s. tanabe etc . for useful discussions during the preparation of this paper .
first of all , we recall the definition of the confluent @xmath0-hypergeometric systems introduced by adolphson @xcite and their important properties . in this paper , we essentially follow the terminology of @xcite and @xcite etc .
let @xmath3 be a finite subset of the lattice @xmath4 .
assume that @xmath0 generates @xmath4 as in @xcite and @xcite .
following @xcite we denote by @xmath20 the convex hull of @xmath21 in @xmath22 . by definition @xmath20 is an @xmath8-dimensional polytope .
let @xmath23 be a parameter vector .
moreover consider the @xmath24 integer matrix @xmath25 whose @xmath26-th column is @xmath27 .
then adolphson s confluent @xmath0-hypergeometric system on @xmath28 associated with the parameter vector @xmath29 is @xmath30 the above @xmath0-hypergeometric system was introduced first by gelfand - kapranov - zelevinsky @xcite under the homogeneous condition on @xmath0 i.e. when there exists a linear functional @xmath31 such that @xmath32 and @xmath33 . in this case , hotta @xcite proved that it is regular holonomic i.e. non - confluent . in @xcite adolphson
does not assume that @xmath0 generates @xmath34 .
however we need this condition to obtain a geometric construction of his confluent @xmath0-hypergeometric systems .
even when @xmath0 does not generate @xmath34 , by a suitable linear coordinate change of @xmath14 we can get an equivalent system for @xmath35 and @xmath36 such that @xmath37 generates @xmath34 . namely our condition is not restrictive at all .
let @xmath38 be the weyl algebra over @xmath39 and consider the differential operators @xmath40 in it .
then the above system is naturally identified with the left @xmath38-module @xmath41 let @xmath42 be the sheaf of differential operators over the
algebraic variety " @xmath39 and define a coherent @xmath42-module by @xmath43 then @xmath44 is the localization of the left @xmath38-module @xmath45 ( see ( * ? ? ?
* proposition 1.4.4 ( ii ) ) etc . ) .
adolphson @xcite proved that @xmath44 is holonomic .
in fact , he proved the following more precise result . [ and ] ( adolphson @xcite , see also @xcite etc . ) for @xmath46 we say that the laurent polynomial @xmath47 is non - degenerate if for any face @xmath1 of @xmath20 not containing the origin @xmath48 we have @xmath49 where we set @xmath50 .
since in the definition above we consider only faces @xmath51 such that @xmath52 and @xmath19 are quasi - homogeneous , our condition is equivalent to the weaker one in @xcite .
see kouchnirenko ( * ? ? ?
* definition 1.19 ) etc .
let @xmath53 be the zariski open subset of @xmath39 consisting of @xmath54 such that the laurent polynomial @xmath47 is non - degenerate .
then adolphson s result in ( * ? ? ?
* lemma 3.3 ) asserts that the holonomic @xmath55-module @xmath44 is an integrable connection on @xmath56 ( i.e. the characteristic variety of @xmath44 is contained in the zero section of the cotangent bundle @xmath57 ) .
now let @xmath58 ( resp .
@xmath59 ) be the underlying complex analytic manifold of @xmath39 ( resp .
@xmath56 ) and consider the holomorphic solution complex @xmath60 of @xmath44 defined by @xmath61 ( see @xcite etc . for the details ) . then by the above adolphson s result , @xmath62 is a local system on @xmath59 .
moreover he proved the following remarkable result .
let @xmath63 be the normalized @xmath8-dimensional volume of @xmath20 i.e. the @xmath64 times of the usual one @xmath65 with respect to the lattice @xmath66 .
[ adl ] ( adolphson ( * ? ? ?
* corollary 5.20 ) ) assume that the parameter vector @xmath67 is semi - nonresonant ( for the definition see @xcite ) .
then the rank of the local system @xmath68 on @xmath59 is equal to @xmath63 .
this is a generalization of the famous result of gelfand - kapranov - zelevinsky in @xcite to the confluent case .
the sections of the local system @xmath69 are called @xmath0-hypergeometric functions ( associated to the parameter @xmath67 ) .
in this section , we review hien s theory of rapid decay homologies invented in @xcite and @xcite . for the theory of twisted homology groups
we refer to aomoto - kita @xcite and pajitnov @xcite etc .
let @xmath70 be a smooth quasi - projective variety of dimension @xmath8 and @xmath71 ( @xmath72 ) an integrable connection on it .
we consider @xmath71 as a left @xmath73-module and set @xmath74.\ ] ] assume that @xmath75 is a smooth projective compactification of @xmath70 such that @xmath76 is a normal crossing divisor and the extension @xmath77 of @xmath78 to @xmath79 admits a good lattice in the sense of sabbah @xcite and mochizuki @xcite .
such a good compactification of @xmath70 for @xmath71 always exists by the fundamental theorem recently established by mochizuki @xcite .
now let @xmath80 be the real oriented blow - up of @xmath81 in @xcite , @xcite and set @xmath82 . recall that @xmath83 induces an isomorphism @xmath84 .
more precisely , for each point @xmath85 by taking a local coordinate @xmath86 on a neighborhood of @xmath87 such that @xmath88 and @xmath89 the morphism @xmath83 is explicitly given by @xmath90 where we set @xmath91 for @xmath92 .
for @xmath93 and a subset @xmath94 denote by @xmath95 the @xmath96-vector space generated by the piecewise smooth maps @xmath97 from the @xmath98-dimensional simplex @xmath99 .
we denote by @xmath100 the sheaf on @xmath101 associated to the presheaf @xmath102 namely @xmath100 is the sheaf of the relative @xmath98-chains on the pair @xmath103 .
now let @xmath104 be the sheaf of horizontal sections of the analytic connection @xmath105 and @xmath106 the inclusion .
then @xmath107 is a local system on @xmath101 .
we define the sheaf @xmath108 of the relative twisted @xmath98-chains on the pair @xmath103 with coefficients in @xmath107 by @xmath109 .
( hien @xcite and @xcite ) a section @xmath110 is called a rapid decay chain if for any point @xmath111 the following condition holds : in a local coordinate @xmath86 on a neighborhood of @xmath87 in @xmath79 such that @xmath88 and @xmath89 by taking a local trivialization @xmath112 with respect to a basis @xmath113 and setting @xmath114 ( @xmath115 ) , for any @xmath116 and @xmath117 there exists @xmath118 such that @xmath119 for any @xmath120 with small @xmath121 . in particular , if @xmath122 we do not impose any condition on @xmath123 . note that this definition does not depend on the local coordinate @xmath86 nor the local trivialization @xmath112 .
we denote by @xmath124 the subsheaf of @xmath108 consisting of rapid decay chains . according to hien @xcite and @xcite
, @xmath124 is a fine sheaf .
then we otain a complex of fine sheaves on @xmath101 : @xmath125.\ ] ] ( hien @xcite and @xcite ) for @xmath126 we set @xmath127 and call it the @xmath98-th rapid decay homology group associated to the integrable connection @xmath78
. in @xcite hien proved that @xmath128 is isomorphic to the dual of the @xmath98-th algebraic de rham cohomology of the dual connection @xmath129 of @xmath78 .
for a different approach to this problem see kashiwara - schapira @xcite . in this paper , we use only some special integrable connections @xmath130 as the following example .
let @xmath131 and @xmath132 , where @xmath133 ( @xmath134 ) is a laurent polynomial and @xmath135 . as usual
we endow @xmath132 with the connection @xmath136 defined by @xmath137 for @xmath138 .
then we have @xmath139 . in this case , to define the rapid decay homology groups @xmath128 we consider ( relative ) twisted chains on which the function @xmath140 decays rapidly at infinity . if @xmath141 and we have an isomorphism @xmath142 ( @xmath143 ) for a possibly multi - valued holomorphic function @xmath144 as the example above , we call @xmath128 the @xmath98-th rapid decay homology group associated to the function @xmath145 . in the special case
where @xmath146 for a meromorphic function @xmath147 on @xmath81 with poles in @xmath148 and a ( possibly multi - valued ) function @xmath149 on @xmath150 such that at each point of @xmath81 there exists a local coordinate @xmath151 satisfying @xmath152 ( @xmath153 ) and @xmath154 , we shall give a purely topological interpretation of @xmath128 . since @xmath79 is a good compactification of @xmath70 for @xmath78 ,
the meromorphic function @xmath147 has no point of indeterminacy on the whole @xmath81 ( see ( * ? ? ?
* section 2.1 ) etc . ) .
by @xmath155 we consider @xmath150 as an open subset of @xmath101 and set @xmath156 let @xmath157 be the irreducible decomposition of @xmath158 . for @xmath159 let @xmath160 be the order of the meromorphic function @xmath147 along @xmath161 .
if @xmath162 we say that the irreducible component @xmath161 is irrelevant .
namely along a relevant component @xmath161 the function @xmath147 has a pole of order @xmath163 .
denote by @xmath164 the union of the irrelevant components of @xmath158 .
then we set @xmath165 .
note that @xmath166 is an open subset of @xmath167 ( i.e. the set of the rapid decay directions of the function @xmath145 in @xmath167 ) .
[ rth ] in the situation as above , we have an isomorphism @xmath168 for any @xmath126 , where the right hand side is the @xmath98-th relative twisted homology group of the pair @xmath169 with coefficients in the rank - one local system @xmath170 on @xmath101 . since the function @xmath171 is single - valued , we have an isomorphism @xmath172 .
first let us consider the case @xmath173 .
locally we may assume that @xmath174 , @xmath175 , @xmath176 and @xmath177 @xmath178 .
let @xmath80 be the real oriented blow - up of @xmath81 along @xmath148 .
in this case we have @xmath179 and @xmath180 . for @xmath181 and sufficiently small @xmath182
we set @xmath183 and @xmath184 .
note that @xmath185 contains all the rapid decay directions of @xmath146 in @xmath167 .
now let us consider the two topological subspaces @xmath186 and @xmath167 of @xmath101 .
we patch them on their intersection @xmath187 and construct a new topological space as follows . by identifying the points of @xmath188 and those of @xmath185 naturally
, we obtain a quotient space @xmath189 of the disjoint union @xmath190 .
recall that @xmath189 is endowed with the strongest topology for which the quotient map @xmath191 is continuous .
note also that @xmath167 is naturally identified with a close subspace of @xmath189 .
we denote the local system on @xmath189 naturally constructed from @xmath107 by the same letter @xmath107 .
for @xmath126 let @xmath192 be the @xmath96-vector space of the twisted ( piecewise smooth ) relative @xmath98-chains on the pair @xmath193 with coefficients in @xmath107 . then by the definition of rapid decay chains , for any @xmath126 we obtain a natural morphism @xmath194 we can easily show that the chain map @xmath195 obtained in this way is a homotopy equivalence .
indeed , its homotopy inverse @xmath196 can be constructed by smooth deformations of chains in @xmath197 in the angular direction @xmath198 .
we can construct them by a smooth vector field on @xmath101 .
hence we obtain an isomorphism @xmath199 for any @xmath126 .
moreover by excision and homotopy , we have an isomorphism @xmath200 for any @xmath126 . combining with
, we obtain an isomorphism @xmath201 for any @xmath126 .
finally let us consider the case @xmath202 .
first we assume that for some @xmath203 we have @xmath204 , @xmath205 , @xmath206 and @xmath207 @xmath208 .
let @xmath80 be the real oriented blow - up of @xmath81 along @xmath148 .
in this case we have @xmath209 . for sufficiently small @xmath182
we define an open subset @xmath185 by @xmath210 for @xmath211 .
then @xmath187 contains all the rapid decay directions of @xmath146 in @xmath167 . as in the case @xmath173 , by smooth deformations of chains and excision etc .
we obtain an isomorphism @xmath212 for any @xmath126 . the general case can be proved similarly by patching local smooth deformations of chains ( smooth vector fields ) as above by a partition of unity .
this completes the proof .
the following lemma will be used in section [ sec:4 ] .
[ ec ] in the situation as above , for @xmath85 let @xmath213 be the number of the relevant irreducible components of @xmath148 passing through @xmath87 .
assume that @xmath214 .
then for a small open neighborhood @xmath215 of @xmath87 in @xmath81 we have @xmath216 by mayer - vietoris exact sequences for relative twisted homology groups , we can reduce the problem to the well - known vanishing of the euler characteristic of the circle @xmath217 . in the sequel
, we consider the more special case where @xmath218 and @xmath130 is an integrable connection on @xmath70 such that @xmath219 for a laurent polynomial @xmath133 ( @xmath134 ) and @xmath135 .
then we can take the projective line @xmath220 to be the good compactification @xmath79 of @xmath218 for @xmath130 . in this case
, we have @xmath221 , where we set @xmath222 and @xmath223 . for the real oriented blow - up @xmath80 of @xmath81
the subset @xmath82 of @xmath101 is a union of two circles @xmath224 ( @xmath225 ) .
moreover the open subset @xmath226 is a union of open intervals in @xmath227 .
let @xmath228 be the newton polytope of @xmath147 i.e. the convex hull of the set @xmath229 in @xmath230
. finally denote by @xmath231 the convex hull of @xmath232 in @xmath230 .
then by proposition [ rth ] and mayer - vietoris exact sequences for relative twisted homology groups we can easily prove the following result .
[ ecl ] in the situation as above , we have \(i ) the dimension of the rapid decay homology group @xmath233 is @xmath234 if @xmath235 and zero otherwise .
\(ii ) assume that @xmath236 $ ] ( resp . @xmath237 $ ] ) for some @xmath238 .
then @xmath226 is a union of @xmath239 open intervals @xmath240 in @xmath241 ( resp . in @xmath242 ) and the first rapid decay homology group @xmath243 has a basis formed by the @xmath239 elements @xmath244 \in h_1^{{{\rm rd}}}(u ; { \mathcal{e } } ) \qquad ( i=1,2 , \ldots , m),\ ] ] where @xmath245 is a @xmath246-dimensional twisted chain with values in @xmath107 starting from a point in @xmath247 and going directly to that in @xmath248 ( here we set @xmath249 ) .
\(iii ) assume that @xmath250 $ ] for some @xmath251 .
then @xmath226 is a union of open intervals @xmath252 in @xmath241 and the ones @xmath253 in @xmath242 .
if moreover the function @xmath149 has a non - trivial monodromy around the origin , then the first rapid decay homology group @xmath243 has a basis formed by the @xmath254 elements @xmath244 \in h_1^{{{\rm rd}}}(u ; { \mathcal{e } } ) \qquad ( i=1,2 , \ldots , m_1)\ ] ] and @xmath255 \in h_1^{{{\rm rd}}}(u ; { \mathcal{e } } ) \qquad ( i=1,2 , \ldots , m_2),\ ] ] where @xmath245 ( resp .
@xmath256 ) is a @xmath246-dimensional twisted chain with values in @xmath107 starting from a point in @xmath247 ( resp .
@xmath257 ) and going directly to that in @xmath248 ( resp .
@xmath258 ) .
in this section we give a geometric construction of adolphson s confluent @xmath0-hypergeometric @xmath2-module @xmath44 and apply it to obtain the integral representations of @xmath0-hypergeometric functions .
let @xmath259 be the dual vector space of @xmath28 , where @xmath260 is the dual coordinate of @xmath261 .
as in @xcite , to @xmath262 we associate a morphism @xmath263 defined by @xmath264 . since we assume here that @xmath0 generates @xmath34 , @xmath26 is an embedding .
let @xmath265 $ ] be the defining ideal of the closure @xmath266 of @xmath267 in @xmath268 .
moreover denote by @xmath269 the weyl algebra over @xmath268 .
then we have a ring isomorphism @xmath270 defined by @xmath271 we call @xmath272 the fourier transform ( see malgrange @xcite etc . for the details ) . via this @xmath272
, the adolphson s system @xmath273 is transformed to the one @xmath274 on @xmath275 .
note that this system has no holomorphic solution in general .
let @xmath276 be the corresponding left @xmath269-module and @xmath277 the coherent @xmath278-module associated to it . by a theorem of hotta @xcite @xmath279
is regular holonomic .
now on the algebraic torus @xmath10 we define a holonomic @xmath280-module @xmath281 by @xmath282 this is an integrable connection on @xmath16 and we have @xmath283.\ ] ] let @xmath284 \in { \mathcal{n}}_{a , c}$ ] and @xmath285 \in { \mathcal{r}}_c$ ] be the canonical generators .
recall that the transfer bimodule @xmath286 has the canonical section @xmath287 .
we define a section @xmath288 of @xmath289 by @xmath290 note that this definition of @xmath288 depends on the coordinates of @xmath268 and @xmath16 .
then we obtain a section @xmath291 of the regular holonomic @xmath278-module @xmath292 defined by @xmath293 .
we can easily check that this section @xmath294 satisfies the system .
hence as in @xcite , we obtain a morphism @xmath295 of left @xmath278-modules which sends the canonical generator @xmath284 \in { \mathcal{n}}_{a , c}$ ] to @xmath294 .
[ nrc ] ( gelfand - kapranov - zelevinsky @xcite ) for a face @xmath1 of @xmath20 containing the origin @xmath48 we denote by @xmath296 the @xmath96-linear span of @xmath1 .
we say that the parameter vector @xmath67 is nonresonant ( with respect to @xmath0 ) if for any face @xmath1 of @xmath20 of codimension @xmath246 such that @xmath297 we have @xmath298 .
recall that if @xmath67 is nonresonant then it is semi - nonresonant in the sense of @xcite .
the following result was proved by saito @xcite and schulze - walther @xcite , @xcite by using commutative algebras . here
we give a geometric proof to it .
[ irr ] assume that the parameter vector @xmath67 is nonresonant .
then the regular holonomic @xmath278-module @xmath299 is irreducible .
note that @xmath300 $ ] is an irreducible perverse sheaf on @xmath301
. then also its minimal extension by the locally closed embedding @xmath26 is irreducible ( see ( * ? ? ?
* corollary 8.2.10 ) etc . ) . as in (
* theorem 3.5 and propositions 3.2 and 4.4 ) it suffices to show that the canonical morphism @xmath302 is a quasi - isomorphism . for this
, we have only to prove the vanishing @xmath303 for any @xmath304 .
note that by the nonresonance of @xmath67 for any @xmath126 and the local system @xmath305 on @xmath301 we have @xmath306 .
let @xmath307 ( resp .
@xmath308 ) be the semigroup ( resp .
the convex cone ) generated by @xmath0 .
then by ( the proof of ) ( * ? ? ?
* chapter 5 , theorem 2.3 ) we have @xmath309)$ ] .
let us define an action of @xmath16 on @xmath310 by @xmath311 for @xmath312 .
then by ( * ? ? ?
* chapter 5 , theorem 2.5 ) there exists a natural bijection between the faces of @xmath313 and the @xmath16-orbits in @xmath266 . in particular ,
if @xmath314 we have @xmath315 and there is nothing to prove . first consider the case where @xmath48 is an appex of @xmath313 and @xmath316 .
if @xmath317 i.e. @xmath318 for some @xmath319 we have @xmath320 .
hence we may assume that @xmath321 from the first .
in this case , @xmath322 is the unique @xmath323-dimensional @xmath16-orbit in @xmath266 which corresponds to @xmath324 . from now on
, we will prove that @xmath325 . by our assumption there
exists a linear function @xmath326 such that @xmath327 and @xmath328 .
we define a real - valued function @xmath329 by @xmath330 where we take @xmath331 large enough so that @xmath332 and its level sets @xmath333 ( @xmath334 ) are smooth .
let @xmath335 be the coefficients of the linear function @xmath336 .
define an action of the multiplicative group @xmath337 on @xmath16 by @xmath338 for @xmath339 .
then we have @xmath340 and hence @xmath341 therefore by the action of @xmath337 on @xmath310 defined by @xmath342 a level set @xmath343 ( @xmath344 ) of @xmath332 is sent to the one @xmath345 .
moreover this action preserves the @xmath16-orbits in @xmath266 .
let @xmath346 be such a @xmath16-orbit
. then all the level sets @xmath343 ( @xmath344 ) of @xmath332 are transversal to @xmath347 , or all are not . but the latter case can not occur by the sard theorem .
then we obtain an isomorphism @xmath348 for any @xmath126 .
next consider the remaining case where @xmath349 for a @xmath16-orbit @xmath347 in @xmath266 such that @xmath350 .
then in a neighborhood of @xmath87 , the variety @xmath266 is a product @xmath351 for an affine toric variety @xmath352 and @xmath353 for some tori @xmath354 .
see ( * ? ? ?
* chapter 5 , theorem 3.1 ) and the proof of ( * ? ?
* theorem 4.9 ) etc . for the details .
moreover for the semigroup @xmath355 generated by a finite subset @xmath356 we have @xmath357 ) \subset w$ ] ( @xmath358 ) .
these varieties @xmath359 are the irreducible components of @xmath360 . for
the explicit construction of @xmath359 see the proof of ( * ? ? ?
* theorem 4.9 ) . by this construction @xmath361 is an appex of the convex cone @xmath362 generated by @xmath363 .
let @xmath364 and @xmath365 be the second projections .
then it follows from the nonresonance of @xmath67 the restriction of @xmath366 to @xmath367 is a non - constant local system .
so we can apply our previous arguments and prove @xmath368 in this case , too .
this completes the proof . by lemma [ irr ] ,
if @xmath67 is nonresonant the non - trivial morphism @xmath369 should be surjective . according to schulze - walther ( * ? ? ?
* corollary 3.8 ) the morphism @xmath369 is also an isomorphism in this case .
let @xmath370 be the inverse of the fourier transform @xmath272 .
then we have an isomorphism @xmath371 of left @xmath38-modules .
the corresponding coherent @xmath55-module @xmath372 can be more geometrically constructed as follows .
let @xmath373 be the canonical pairing defined by @xmath374 and @xmath375 ( resp .
@xmath376 ) the first ( resp .
second ) projection .
then we have the following theorem due to katz - laumon @xcite . [ kal ] ( katz - laumon @xcite ) in the situation as above , we have an isomorphism @xmath377 where @xmath378 is the integrable connection associated to @xmath379 ( see @xcite etc . ) . in the same way , we have @xmath380 from now on , we assume that @xmath67 is nonresonant .
then by lemma [ irr ] we obtain surjective morphisms @xmath381 and @xmath382 .
hence we obtain a surjective morphism @xmath383 of left @xmath55-modules .
let @xmath384 be the function defined by @xmath385 and @xmath386 ( resp .
@xmath387 ) the first ( resp .
second ) projection .
then by the base change theorem ( * ? ? ?
* theorem 1.7.3 and corollary 1.7.5 ) , we have the isomorphism @xmath388 namely @xmath389 is the direct image of the integrable connection @xmath390 on @xmath391 by @xmath392 . define a function @xmath393 by @xmath394 then by the results of hien - roucairol @xcite the holomorphic solution complex @xmath395 of @xmath389 is expressed by the rapid decay homology groups associated the function @xmath145 . indeed , for @xmath396 let @xmath397 ( resp .
@xmath398 ) be the restriction of the connection @xmath399 ( resp .
the function @xmath145 ) to @xmath400 .
namely we set @xmath401 for @xmath402 .
then @xmath403 and for the dual connection @xmath404 of @xmath397 we have @xmath405 moreover for any @xmath126 , by proposition [ rth ] ( see also the proof of theorem [ main ] below ) the rapid decay homology groups @xmath406 associated to the integrable connections @xmath407 ( or to the functions @xmath408 ) are isomorphic to each other and define a local system @xmath409 on @xmath59 .
see @xcite for the details .
the following result is essentially due to hien - roucairol @xcite .
[ har ] ( hien - roucairol @xcite ) in the situation as above , for any @xmath126 we have an isomorphism @xmath410 of local systems on @xmath59 . in ( * ? ? ?
* section 3 ) adolphson proved that @xmath44 is an integrable connection on @xmath56 .
then by the surjective morphism @xmath411 we find that @xmath389 is also an integrable connection on @xmath56 .
this in particular implies that for any @xmath412 we have @xmath413 .
hence we get @xmath414 it follows also from the surjection @xmath415 that we have an injection @xmath416 by using the generator @xmath417 \in { \mathcal{m}}_{a , c}= { \mathcal{d}}_{x } / \left(\sum_{1 \leq i \leq
n } { \mathcal{d}}_{x } z_{i , c } + \sum_{\mu \in { \rm ker } a \cap { \mathbb{z}}^{n } } { \mathcal{d}}_{x } \square_{\mu } \right)\ ] ] of @xmath44 we regard @xmath418 as a subsheaf of @xmath419 .
then we have the following result . [ main ]
assume that the parameter vector @xmath67 is nonresonant .
then the morphism @xmath420 induces an isomorphism @xmath421 of local systems on @xmath59 .
moreover this isomorphism is given by the integral @xmath422 where for a continuous family @xmath423 of rapid decay @xmath8-cycles in @xmath424 and @xmath425 we denote by @xmath9 its restriction @xmath426 to @xmath427 .
note that this integral representation of the confluent @xmath0-hypergeometric functions @xmath418 coincides with the one in adolphson ( * ? ? ? * equation ( 2.6 ) ) .
recall that the sheaf @xmath428 is a local system on @xmath59 .
moreover by ( * ? ? ?
* corollary 5.20 ) its rank is @xmath429 .
so it suffices to show that for any @xmath425 the dimension of the @xmath8-th rapid decay homology group @xmath430 is also @xmath429 .
let @xmath431 be the rapid decay euler characteristic .
then by we have only to prove the equality @xmath432 let @xmath433 be the dual fan of @xmath20 in @xmath14 and @xmath434 its smooth subdivision
. denote by @xmath435 the smooth toric variety associated to the fan @xmath434 .
then @xmath435 is a smooth compactification of @xmath436 such that @xmath437 is a normal crossing divisor . by using the non - degeneracy of the laurent polynomial @xmath17 , as in (
* section 3 ) we can construct a complex blow - up @xmath438 of @xmath435 such that the meromorphic extension of @xmath19 to it has no point of indeterminacy .
for the reader s convenience , we briefly recall the construction of @xmath79 . recall that @xmath16 acts on @xmath435 and the @xmath16-orbits are parametrized by the cones in @xmath434 . for a cone
@xmath439 we denote by @xmath440 the corresponding @xmath16-orbit .
let @xmath441 be the rays i.e. the one - dimensional cones in @xmath434 . by using the primitive vectors @xmath442 on @xmath443 we set @xmath444 we renumber @xmath445 so that @xmath446 if and only if @xmath447 for some @xmath448 .
then for any @xmath447 the meromorphic extension of @xmath19 to @xmath435 has a pole of order @xmath449 along the toric divisor @xmath450 . by the non - degeneracy of @xmath19
the hypersurface @xmath451 intersects @xmath452 transversally for any subset @xmath453 such that @xmath454 ( see definition [ and ] ) .
the meromorphic extension of @xmath19 has points of indeterminacy in @xmath455 .
figure 1 .
first we construct a tower of @xmath456 codimension - two blow - ups over @xmath457 ( see ( * ? ? ?
* section 3 ) and ( * ? ? ? * section 3 and lemma 4.9 ) for the details ) .
then the indeterminacy of @xmath19 over @xmath458 is eliminated . by repeating this construction also over ( the proper transforms of ) @xmath459
we finally obtain the desired proper morphism @xmath460 of @xmath435 as the figure below .
figure 2 .
now we can use this smooth compactification @xmath79 of @xmath461 and the normal crossing divisor @xmath462 in it to define the rapid decay homology groups associated to the function @xmath463 ( see ( * ? ? ? * section 2.1 ) etc . ) . in the figure of @xmath79 above
, the dotted curves stand for the irrelevant components of @xmath158 .
note that for any @xmath447 the inverse image of @xmath464 by the morphism @xmath465 is a union of @xmath466-bundles on it and only the last one among them is irrelevant .
let @xmath80 be the real oriented blow - up of @xmath81 along @xmath148 and set @xmath467 . then , as in section [ sec:3 ] we define the rapid decay homology groups @xmath468 by using @xmath80 , @xmath167 and @xmath469 etc . by defining @xmath226
etc . as in proposition
[ rth ] , for the local system @xmath470 on @xmath471 we obtain isomorphisms @xmath472 for @xmath447 we define a face @xmath473 of @xmath20 by @xmath474 we call it the supporting face of @xmath443 in @xmath20 .
denote by @xmath475 the normalized @xmath476-dimensional volume @xmath477 of @xmath478 .
then the @xmath478-part @xmath479 of @xmath19 is naturally identified with the defining ( laurent ) polynomial of the hypersurface @xmath480 in @xmath481 .
moreover by the bernstein - khovanskii - kouchnirenko theorem ( see @xcite etc . ) its euler characteristic is equal to @xmath482 .
note that we have @xmath483 .
then by applying lemmas [ ec ] and [ ecl ] and mayer - vietoris exact sequences for relative twisted homology groups to the geometric situation in figure @xmath484 above , we can easily calculate the rapid decay euler characteristic @xmath485 and prove the equality .
this completes the proof of the isomorphism .
let us prove the remaining assertion . denote the distinguished section @xmath486 of the integrable connection @xmath487 by @xmath488 .
let @xmath489 be the relative algebraic de rham complex of @xmath399 associated to the morphism @xmath490 .
then we have an isomorphism @xmath491 for a relative @xmath8-form @xmath492 denote by @xmath493 the section of @xmath389 which corresponds to the cohomology class @xmath494 \in h^n \ { ( q_1)_*(\omega^{\cdot}_{x \times t / x } \otimes_{{\mathcal{o}}_{x \times t } } { \mathcal{k } } ) \}$ ] by the above isomorphism . according to the result of @xcite , by the isomorphism @xmath495 of local systems on @xmath59 , a family of rapid decay cycles @xmath496
is sent to the section @xmath497\ ] ] ( @xmath498 ) of @xmath499 .
then the remaining assertion follows from the lemma below .
this completes the proof .
[ tran ] when @xmath18 the irrelevant components of @xmath158 in the proof above are the last @xmath466-bundles on @xmath464 ( @xmath500 ) . by the construction of the morphism @xmath460
we can easily see that for any @xmath501 the hypersurface @xmath502 intersects them transversally . by the morphism
@xmath503 the canonical section @xmath504 \in { \mathcal{m}}_{a , c}$ ] is sent to the cohomology class @xmath505 .
first note that the morphism @xmath506 sends the canonical generator @xmath504
\in { \mathcal{m}}_{a , c}(x)$ ] to @xmath507 . on the other hand ,
by we have an isomorphism @xmath508.\ ] ] then by malgrange s simple proof @xcite of theorem [ kal ] , via this isomorphism the section @xmath509 corresponds to the cohomology class @xmath510.\ ] ] let @xmath511 be the embedding induced by @xmath26 . by the isomorphism @xmath512\ ] ] the above cohomology class corresponds to the one @xmath513,\ ] ] where the section @xmath514 is defined similarly to @xmath515 . then it suffices to show that via the isomorphism @xmath516 the cohomology class @xmath517 is sent to the one @xmath518 $ ] in @xmath519 since @xmath39 and @xmath520 are affine , we have only to prove that via the isomorphism @xmath521 the cohomology class @xmath522\ ] ] is sent to the one @xmath523 $ ] .
indeed , we have isomorphisms @xmath524 by which the element @xmath525 $ ] is sent to the one @xmath526 $ ] .
let @xmath527 be a free resolution of the left @xmath528-module @xmath399 .
since @xmath391 is affine , we obtain a surjective homomorphism @xmath529 and can take a lift @xmath530 of @xmath531 .
moreover by the flatness of the right @xmath280-module @xmath532 and the well - known formula @xmath533 there exists an isomorphism @xmath534 by which @xmath526 $ ] is sent to @xmath535 $ ] .
similarly , by the isomorphism @xmath536 the element @xmath535 $ ] is sent to @xmath537 $ ] .
this completes the proof . as a corollary of theorem [ main ]
, we recover the following saito and schulze - walther s geometric ( functorial ) construction of adolphson s confluent @xmath0-hypergeometric @xmath2-module @xmath538 on @xmath539 .
( saito @xcite and schulze - walther @xcite , @xcite ) assume that the parameter vector @xmath67 is nonresonant .
then we have an isomorphism @xmath540 of integrable connections on @xmath56 .
in particular , @xmath538 is an irreducible connection there .
this result was first obtained in saito @xcite and schulze - walther @xcite , @xcite by using totally different methods .
in fact , they proved moreover that we have an isomorphism @xmath540 on the whole @xmath39 . since @xmath279 is regular holonomic by a theorem of hotta @xcite , it is also regular at infinity in the sense of daia @xcite .
then by using the fourier - sato transforms ( see @xcite and @xcite etc . ) , we can apply the main theorem of daia @xcite to get a topological construction of the sheaf of the confluent @xmath0-hypergeometric functions @xmath418 .
this construction is valid even when the parameter @xmath67 is not nonresonant .
assume that @xmath173 and @xmath541 .
\(i ) if @xmath542 our integral representation of the @xmath0-hypergeometric functions @xmath543 on @xmath544 is @xmath545 by the restriction of @xmath543 by the injective map @xmath546 , @xmath547 we obtain the classical bessel function @xmath548 for the parameter @xmath549 .
\(ii ) if @xmath550 our integral representation of the @xmath0-hypergeometric functions @xmath543 on @xmath544 is @xmath551 by the restriction of @xmath543 by the injective map @xmath546 , @xmath552 we obtain the classical airy function @xmath553 for @xmath554 .
in this section , assuming the condition @xmath555 we construct natural bases of the rapid decay homology groups @xmath556 and apply them to prove a formula for the asymptotic expansions at infinity of adolphson s confluent @xmath0-hypergeometric functions . for the construction of them , we first prove that the following subset @xmath557 of @xmath56 is open dense in @xmath558 .
we define a subset @xmath557 of @xmath559 by : @xmath560 @xmath561 @xmath396 and the laurent polynomial @xmath47 has only non - degenerate ( morse ) critical points in @xmath562 .
it is clear that @xmath563 is stable by the multiplication of @xmath564 ( i.e. homothecy ) on @xmath558 .
let @xmath565 be elements of @xmath0 such that @xmath566 is a basis of the vector space @xmath14 . by our assumption that @xmath0 generates @xmath34
, we can take such elements of @xmath0 .
let @xmath567 be a laurent polynomial with support in @xmath262 on @xmath562 .
assume that @xmath147 is non - degenerate i.e. @xmath568 .
then for generic @xmath569 the perturbation @xmath570 of @xmath147 is non - degenerate and has only non - degenerate ( morse ) critical points in @xmath16 .
it is clear that @xmath571 is non - degenerate for generic @xmath572 ( see for example ( * ? ? ?
* lemma 5.2 ) etc . ) .
let @xmath573 be the dual basis of @xmath574 and set @xmath575 note that for @xmath576 we have @xmath577 then we can easily prove the equality @xmath578 where @xmath579 is an invertible matrix defined by @xmath580 .
hence we obtain @xmath581 moreover degenerate critical points of @xmath571 in @xmath16 correspond to critical points @xmath312 of the map @xmath582 such that @xmath583 ( @xmath584 ) . by the bertini - sard theorem ,
generic @xmath569 are not such critical values .
this implies that for generic @xmath572 the laurent polynomial @xmath571 has no degenerate critical point .
this completes the proof .
[ opd ] the subset @xmath557 of @xmath56 is open dense in @xmath558 and stable by the multiplication of @xmath564 ( i.e. homothecy ) on @xmath558 .
[ ncp ] assume that @xmath555 .
then for any @xmath560 the laurent polynomial @xmath17 has exactly @xmath585 non - degenerate ( morse ) critical points in @xmath16 .
let us fix @xmath560 and set @xmath586 . by an invertible matrix @xmath587 we define new laurent polynomials @xmath588 on @xmath16 by @xmath589 by our assumption @xmath555 , taking
sufficiently generic @xmath590 we may assume that all the newton polytopes of @xmath588 are equal to @xmath20 .
then for any face @xmath51 of @xmath20 the set @xmath591 coincides with that of the critical points of @xmath592 in @xmath16 . in this correspondence for the special case @xmath593 , multiple roots of the equation @xmath594 in @xmath16 correspond to degenerate critical points of @xmath595 .
but by our assumption @xmath560 there is no such point in @xmath16 .
moreover by the non - degeneracy of @xmath147 ( @xmath596 ) , for any face @xmath51 of @xmath20 such that @xmath597 ( i.e. @xmath598 when @xmath18 ) we have @xmath599 this means that the ( @xmath323-dimensional ) subvariety @xmath600 of @xmath16 is a non - degenerate complete intersection ( for the definition , see ( * ? ? ?
* definition 2.7 ) and @xcite etc . ) . then by bernstein s theorem its cardinality is equal to @xmath585 . from now on ,
assuming the condition @xmath555 , for any @xmath560 we construct a natural basis of the rapid decay homology group @xmath556 by using the ( relative ) twisted morse theory for the function @xmath601 .
for the twisted morse theory and its applications to period integrals , we refer to aomoto - kita @xcite , pajitnov @xcite and pham @xcite etc .
our construction of the basis is similar to the ones of dubrovin @xcite and tanabe - ueda @xcite etc . in the untwisted case .
note that by our assumption @xmath555 any parameter vector @xmath67 is nonresonant . for @xmath560 let @xmath602 ( @xmath603 ) be the non - degenerate ( morse ) critical points of the laurent polynomial @xmath47 in proposition [ ncp ] . by the cauchy - riemann equation , they are also non - degenerate ( morse ) critical points of the real - valued function @xmath601 .
we can observe this fact more explicitly by taking a holomorphic morse coordinate around each @xmath602 as follows .
for a fixed @xmath603 let @xmath604 , @xmath605 ( @xmath606 ) be a holomorphic morse coordinate for @xmath19 around its critical point @xmath602 such that @xmath607 in a neighborhood of @xmath602 .
since we have @xmath608 we regard the smooth submanifold @xmath609 in it as the stable manifold of the gradient flow of the morse function @xmath601 in a neighborhood of @xmath610 and denote it by @xmath611 . by shrinking @xmath611 if necessary
, we may assume that @xmath612 is homeomorphic to the @xmath8-dimensional disk . for @xmath603 let @xmath613 be the ray in @xmath96 defined by @xmath614 namely @xmath615 emanates from the critical value @xmath616 of @xmath19 and goes to the left in the complex plane @xmath96 so that we have @xmath617 along it . by shrinking the stable manifold @xmath611 if necessary
, we may assume also that the image of @xmath618 by the map @xmath619 is the closed interval @xmath620 in @xmath615 for some @xmath182 and @xmath621 is just the one point @xmath622 in @xmath615 .
we drag @xmath623 over the complement of @xmath624 in @xmath615 to construct a tube @xmath625 \times s^{n-1}$ ] in @xmath301
. then @xmath626 is a singular @xmath8-chain in @xmath301 whose boundary in the real oriented blow - up @xmath101 is contained in the subset @xmath226 .
we thus constructed a rapid decay @xmath8-cycle @xmath245 in @xmath301 for the function @xmath627 such that @xmath628 [ brh ] in the situation as above ( i.e. @xmath555 and @xmath560 etc . ) , the elements @xmath629 , [ \gamma_2 ] , \ldots , [ \gamma_{{{\rm vol}}_{{\mathbb{z } } } ( \delta ) } ] \in ( { \mathcal{h}}_{n}^{{{\rm rd}}})_z \simeq h_n^{{{\rm rd } } } ( u_z ; { \mathcal{k}}_z^ * ) $ ] form a basis of the rapid decay homology group @xmath630 .
we use the notations in proposition [ rth ] and the proof of theorem [ main ] by setting @xmath631 then by proposition [ rth ] , for @xmath632 we have an isomorphism @xmath633 note that by we have @xmath634 for @xmath635 we define an open subset @xmath636 of @xmath301 by @xmath637 then by remark [ tran ] for any @xmath635 the closure of @xmath638 in @xmath79 intersects each irrelevant divisor @xmath639 transversally .
this implies that for any @xmath126 and @xmath640 we have @xmath641 moreover for any @xmath126 and @xmath642 we have an isomorphism @xmath643 now let @xmath644 be the critical values of @xmath601 .
then by remark [ tran ] for any @xmath126 and @xmath645 such that @xmath646 , @xmath647 \cap \ { t_1 , t_2 , \ldots , t_r \ } = \emptyset$ ] we have a natural isomorphism @xmath648 for @xmath649 let @xmath650 be the critical points of @xmath651 such that @xmath652 .
then , for sufficiently small @xmath653 we obtain a short exact sequence @xmath654 in view of , , and the fact @xmath655 .
moreover the canonical generators @xmath656 \in h_n ( \overline{s_{i_q } } , \partial s_{i_q } ; \iota _ * { \mathcal{l } } ) \simeq { \mathbb{c}}$ ] can be lifted to the elements @xmath657 $ ] of @xmath658 .
this implies that @xmath629 , [ \gamma_2 ] , \ldots , [ \gamma_{{{\rm vol}}_{{\mathbb{z } } } ( \delta ) } ] \in h_n(t^{{{\rm an } } } \cup q , q ; \iota _ * { \mathcal{l}})$ ] form a basis of @xmath659 .
note that for a connected open neighborhood @xmath215 of the point @xmath261 in @xmath660 the basis @xmath629 , \ldots , [ \gamma_{{{\rm vol}}_{{\mathbb{z } } } ( \delta ) } ] \in ( { \mathcal{h}}_{n}^{{{\rm rd}}})_z$ ] constructed in theorem [ brh ] can be naturally extended to a family of the bases @xmath661 , \ldots , [ \gamma^w_{{{\rm vol}}_{{\mathbb{z } } } ( \delta ) } ] \in ( { \mathcal{h}}_{n}^{{{\rm rd}}})_w$ ] ( @xmath662 ) i.e. a basis of the local system @xmath663 on @xmath215 .
we can extend it so that @xmath664 is stable by the multiplication of the group @xmath337 on @xmath665 and the rapid decay @xmath8-cycles @xmath666 ( @xmath662 ) satisfy the conditions @xmath667 where @xmath668 is the stable manifold of the gradient flow of @xmath669 passing through its @xmath670-th non - degenerate critical point @xmath671 . for @xmath603
we define a confluent @xmath0-hypergeometric function @xmath672 on @xmath664 by @xmath673 for @xmath662 . then by applying the higher - dimensional saddle point ( steepest descent ) method
to holomorphic morse coordinates around the critical points of @xmath674 ( @xmath662 ) in @xmath301 we obtain the following result . for @xmath675 let @xmath676 be the open sector in @xmath96 defined by @xmath677 . by taking a sufficiently small @xmath675 such that @xmath678 for any @xmath679 we set @xmath680 . [ ase ] in the situation as above ( i.e. @xmath18 ) , if @xmath675 is sufficiently small , for any @xmath603 and @xmath681 we have an asymptotic expansion : @xmath682 as @xmath683 in the sector @xmath684 , where @xmath685 are functions of @xmath261 and @xmath686
is the hessian of @xmath19 at @xmath687 .
first , it is clear that for any @xmath681 the critical points of the function @xmath688 are @xmath689 ( @xmath690 ) .
fix @xmath603 .
let @xmath604 , @xmath605 ( @xmath691 ) be the holomorphic morse coordinate for the function @xmath19 around its @xmath670-th critical point @xmath610 such that @xmath692 .
for sufficiently small @xmath182 we define an open neighborhood @xmath693 of @xmath610 by @xmath694 and set @xmath695 in it .
then @xmath696 is the stable manifold of the gradient flow of the function @xmath697 passing through its non - degenerate critical point @xmath610 . for @xmath698 ( @xmath699 )
we set @xmath700 .
then the laurent polynomial @xmath701 can be rewritten as @xmath702 ( @xmath703 ) we see also that the subset @xmath704 of @xmath693 is the stable manifold of the gradient flow of @xmath705 through @xmath610 . by our construction of the rapid decay @xmath8-cycle @xmath706 we may assume that @xmath707 and @xmath708 for any @xmath709 .
we may assume also that for any @xmath710 such that @xmath711 we have @xmath712 .
this implies that ( if @xmath675 is sufficiently small ) there exists a positive real numbers @xmath713 such that @xmath714 for any @xmath681 .
then we have @xmath715 for any @xmath681 satisfying @xmath716 .
hence , to prove the theorem , it suffices to calculate the asymptotic expansion of the integral @xmath717 as @xmath683 in the sector @xmath684 . for the morse coordinate @xmath604 of @xmath19
we can easily show @xmath718 also by using the coordinate @xmath604 set @xmath719 and let @xmath720 be its taylor expansion at @xmath721 i.e. @xmath722 .
then by we obtain @xmath723 now the restriction of the @xmath8-form @xmath724 to the stable manifold @xmath725 has the following form : @xmath726 d \eta_1^{\prime } \wedge \cdots \wedge d \eta_n^{\prime}. \end{aligned}\ ] ] for any @xmath727 we can show that the integral of the @xmath8-form @xmath728 over the whole @xmath729 is equal to @xmath730 by setting @xmath731 . note that the integral @xmath732 is zero if @xmath733 is odd for some @xmath584 . as the previous part of this proof , we can show also that there exists @xmath734 such that @xmath735 for any @xmath681 satisfying @xmath716 .
then the result follows immediately from .
this completes the proof .
if @xmath560 and the critical point @xmath689 of @xmath601 in @xmath301 is given , by using the holomorphic morse coordinate in the proof above we can calculate also the coefficients @xmath736 explicitly .
for the point @xmath560 let @xmath737 be the complex line in @xmath665 passing through @xmath560 . then by theorems [ brh ] and [ ase ] we can observe stokes phenomena for the restrictions @xmath738 of the functions @xmath672 ( @xmath690 ) to the line @xmath739 . indeed , by theorem [ ase ] the dominance ordering of the functions @xmath738 at infinity ( i.e. where @xmath716 ) changes as @xmath740 increases .
more precisely , the asymptotic expansions at infinity of the restrictions of the @xmath0-hypergeometric functions to @xmath739 may jump at the stokes lines : @xmath741 = 0 \right\ } \qquad ( i \not= j).\ ] ] it would be an interesting problem to determine the stokes multipliers in this case .
in this section , we shall construct a natural basis of the rapid decay homology group @xmath742 in the two - dimensional case i.e. @xmath743 .
first , we prepare some elementary results on relative twisted homology groups . set @xmath744 and let @xmath745 be the meromorphic function on @xmath81 defined by @xmath746 let @xmath747 be the real oriented blow - up of @xmath81 along the normal crossing divisor @xmath748 and set @xmath749 and @xmath750 . by the inclusion map @xmath751 we consider @xmath752 as an open subset of @xmath753 and set @xmath754 and @xmath755 .
finally let @xmath756 be the local system of rank one on @xmath752 defined by @xmath757 then by homotopy and lemma [ ecl ] we obtain the following lemma .
[ 2-dim-1 ] \(i ) for @xmath758 set @xmath759 then for any @xmath126 the natural morphism @xmath760 is an isomorphism .
\(ii ) assume that @xmath761 satisfies the condition @xmath762 .
then we have @xmath763 for any @xmath126 .
the assertion ( i ) can be easily shown by homotopy . we will prove ( ii ) .
let @xmath217 be the unit circle @xmath764 in @xmath765 .
then by ( i ) and homotopy we have an isomorphism @xmath766 for any @xmath126 .
let us take the base point @xmath767 of @xmath217 .
then by lemma [ ecl ] we have @xmath768 for @xmath769 and there exists a natural basis @xmath629 , \ldots , [ \gamma_{m_2}]$ ] of @xmath770 .
let @xmath771 be the linear automorphism , i.e. the monodromy of @xmath772 induced by the ( clockwise ) rotation along the circle @xmath217 . by the matrix representation of @xmath773 with respect to the basis @xmath629 , \ldots , [ \gamma_{m_2}]$ ] we see that the eigenvalues of @xmath773 are contained in the set @xmath774 \right\}.\ ] ] in particular , our assumption @xmath762 implies that @xmath775 is an automorphism of @xmath772 .
now for @xmath776 we define two arcs @xmath777 in @xmath217 by @xmath778 then @xmath779 . by the mayer - vietoris exact sequence for relative twisted homology groups associated to the open covering @xmath780 of @xmath781 we can calculate @xmath782 ( @xmath126 ) from @xmath783 ( @xmath126 ) .
then the assertion ( ii ) follows from the invertibility of @xmath775 .
next consider the meromorphic function @xmath784 on @xmath785 defined by @xmath786 where @xmath787 are distinct non - zero complex numbers .
let @xmath788 be the real oriented blow - up of @xmath81 along @xmath789 and define @xmath790 , @xmath791 , @xmath792 and @xmath793 as above .
moreover let @xmath794 be the local system of rank one on @xmath795 defined by @xmath796 then by the proof of lemma [ 2-dim-1 ] ( ii ) and mayer - vietoris exact sequences for relative twisted homology groups we obtain the following proposition .
[ 2-dim-2 ] \(i ) for @xmath758 set @xmath797 then for any @xmath126 the natural morphism @xmath798 is an isomorphism .
\(ii ) assume that @xmath799 and @xmath761 , @xmath800 satisfy the conditions @xmath762 and @xmath801 for any @xmath802 .
then we have @xmath803 and can explicitly construct a basis of the vector space @xmath804 over @xmath96 . in the special but important case
where @xmath214 and @xmath805 , we can construct the basis of @xmath804 in proposition [ 2-dim-2 ] ( ii ) very elegantly as follows . by homotopy
we may assume that @xmath806 ( @xmath802 ) from the first .
for @xmath802 let @xmath807 be the arc in the unit circle @xmath217 between the two points @xmath808 , where we set @xmath809 . for sufficiently small @xmath182
let @xmath810 be the boundary of the set @xmath811 and denote the central point of the arc @xmath812 by @xmath813 .
we regard @xmath814 as the base point of the one - dimensional complex @xmath810 .
then by lemma [ ecl ] we have @xmath815 for @xmath769 and there exists a natural basis @xmath816 , \ldots , [ \gamma_{j m_2}]$ ] of @xmath817 .
moreover by the proof of lemma [ 2-dim-1 ] ( ii ) and mayer - vietoris exact sequences we obtain @xmath818 for @xmath819 .
now we start from the base point @xmath814 , draw the figure @xmath820 along @xmath810 and come back to the same place @xmath814 . by our assumption @xmath821 we can drag the twisted chains @xmath822 over the figure @xmath820 ( keeping their end points in the rapid decay direction @xmath823 of @xmath824 ) to get a basis @xmath825 , \ldots , [ \delta_{j m_2}]$ ] of @xmath826 .
on the other hand , by proposition [ 2-dim-2 ] ( i ) and homotopy there exists an isomorphism @xmath827 for any @xmath126 .
moreover it follows from our assumption @xmath828 that we have @xmath829 for any @xmath802 and @xmath126 . hence by the mayer - vietoris exact sequences associated to the covering @xmath830 of @xmath831 we obtain the following result .
[ 2-dim - new ] assume that @xmath214 , @xmath805 and @xmath761 , @xmath800 satisfy the condition @xmath832 for @xmath802 .
then the elements @xmath825 , \ldots , [ \delta_{j m_2 } ] \in h_2(u_1^{{{\rm an } } } \cup q_1 , q_1 ; ( \iota_1 ) _ * { \mathcal{l}}_1 ) $ ] ( @xmath802 ) constructed above are linearly independent over @xmath96 and form a basis of @xmath804 .
now let us return to our original situation in sections [ sec:4 ] and [ sec:5 ] in the two - dimensional case .
for @xmath396 we define @xmath833 in the real oriented blow - up @xmath80 of @xmath834 as in the proof of theorem [ main ] . for the local system @xmath835 on @xmath301
we shall construct a basis of the rapid decay homology group @xmath836 . by abuse of notations , for an open subset @xmath360 of @xmath301 and @xmath126
we set @xmath837 for short .
recall that @xmath434 is a smooth subdivision of the dual fan of @xmath838 and @xmath839 are the rays i.e. the one - dimensional cones in @xmath434 which correspond to the relevant divisors @xmath840 in @xmath841 .
we renumber @xmath842 in the clockwise order so that we have @xmath843 for any @xmath844 . by the primitive vector @xmath845 on @xmath443 the order @xmath446 of the pole of @xmath846 along @xmath161
is explicitly given by @xmath847 for @xmath447 we set @xmath848 then at each point of @xmath849 there exists a local coordinate system @xmath850 of @xmath435 such that @xmath851 and the function @xmath852 has the form @xmath853 .
namely the function @xmath852 has the order @xmath854 along @xmath161 . by the non - degeneracy of @xmath19 the complex curve @xmath855
intersects each relevant divisor @xmath161 transversally .
set @xmath856 and @xmath857 . by our construction of the complex blow - up @xmath858 of @xmath435 , the fiber of the point
@xmath859 is a union @xmath860 of exceptional divisors @xmath861 along which @xmath19 has a pole of order @xmath862 .
see figure 3 below .
moreover the order of the function @xmath852 along @xmath863 is @xmath864 for any @xmath865
. let @xmath866 be the one - dimensional @xmath16-orbit in @xmath435 which corresponds to @xmath443 and denote by the same letter @xmath867 its strict transform in the blow - up @xmath868 .
assume that @xmath869 ( @xmath447 ) and @xmath870 ( @xmath844 ) .
then by propositions [ 2-dim-2 ] ( ii ) and [ 2-dim - new ] there exists a sufficiently small neighborhood @xmath871 of @xmath872 in @xmath81 such that for its open subset @xmath873 we have @xmath874 furthermore we can explicitly construct a basis @xmath875 ( @xmath876 , @xmath865 ) of the vector space @xmath877 over @xmath96 .
figure 3 .
[ 2-dim-3 ] assume that @xmath878 is nonresonant , @xmath869 ( @xmath447 ) and @xmath870 ( @xmath844 ) .
then the natural morphisms @xmath879 are injective and induce an isomorphism @xmath880 in particular , the cycles @xmath881 ( @xmath447 , @xmath876 , @xmath865 ) form a basis of the vector space @xmath882 over @xmath96 . by the repeated use of lemma [ 2-dim-1 ] ( i ) and homotopy , we can find a small neighborhood @xmath883 of @xmath884 in @xmath81 containing @xmath871 such that for its open subset @xmath885 the natural morphism @xmath886 is an isomorphism for any @xmath126 . by our assumption
@xmath887 we have the vanishing of the usual twisted homology group @xmath888 for any @xmath126 .
similarly we obtain the vanishings of @xmath889 etc .
then the assertion can be proved by patching these results with the help of lemma [ 2-dim-1 ] ( i ) and mayer - vietoris exact sequences of relative twisted homology groups .
let @xmath473 be the supporting face of @xmath443 in @xmath20 .
then the lattice length of @xmath478 is equal to @xmath475 and we have the equality @xmath890 as expected from the result of theorem [ 2-dim-3 ] .
in this section , we shall extend the construction of the basis of the rapid decay homology group @xmath742 for @xmath743 in section [ sec:6 ] to higher - dimensional cases .
let @xmath891 be a @xmath892-dimensional algebraic torus and @xmath893 a laurent polynomial on it whose newton polytope @xmath894 is @xmath892-dimensional .
we assume that @xmath745 is non - degenerate in the sense of kouchnirenko @xcite .
namely we impose the condition in definition [ and ] for any face @xmath895 of @xmath896 .
[ h - dim-1 ] in the situation as above , for generic @xmath897 the ( possibly multi - valued ) function @xmath898 on @xmath899 has exactly @xmath900 critical points in @xmath899 and all of them are non - degenerate ( i.e. of morse type ) and contained in @xmath901 . for @xmath902 set @xmath903 .
then for @xmath904 we have @xmath905 since the hypersurface @xmath906 is smooth , by all the critical points of the function @xmath907 are contained in @xmath908 .
set @xmath909 .
then by bernstein s theorem we have @xmath910 if the newton polytopes @xmath911 of the laurent polynomials @xmath912 ( @xmath902 ) are equal to @xmath896 and the subvariety @xmath913 of @xmath899 is a non - degenerate complete intersection ( see @xcite etc . ) . from now
on , we will show that these two conditions are satisfied for generic @xmath914 .
first of all , it is clear that @xmath915 ( @xmath902 ) for generic @xmath914 .
note that @xmath904 is in @xmath916 if and only if @xmath917 and @xmath918 that is , @xmath917 is sent to the point @xmath914 by the map @xmath919 defined by @xmath920 by the bertini - sard theorem generic @xmath914 are regular values of this map .
if @xmath914 is such a point , we can easily show that @xmath921 for any @xmath922 .
now let @xmath923 be a proper face of @xmath896 .
then for generic @xmath914 we have @xmath924 indeed , let us assume the converse .
then the first projection from the variety @xmath925 to @xmath926 is dominant .
moreover by @xmath927 this variety is quasi - homogeneous with respect to the second variables @xmath928 .
in particular , its dimension is greater than @xmath892 .
then by considering the second projection from it to @xmath899 , we find that there exist @xmath904 and @xmath929 such that @xmath930 for any @xmath902
. namely we have @xmath931 and @xmath932 comparing with for @xmath902 such that @xmath933 , we obtain @xmath934 for the point @xmath904 .
this contradicts our assumption that the laurent polynomial @xmath745 is non - degenerate .
we thus proved that the subvariety @xmath913 of @xmath899 is a non - degenerate complete intersection and its cardinality is @xmath900 for generic @xmath914 .
let us fix such a point @xmath914 . recall that @xmath935 is the set of the critical points of the function @xmath936 in @xmath908 .
at such a critical point @xmath917 we have @xmath937 since @xmath938 , we obtain @xmath939 namely all the critical points of the function @xmath936 are non - degenerate .
let @xmath745 be as above and @xmath756 a local system of rank one on @xmath940 defined by @xmath941 from now on , we will calculate the twisted homology groups @xmath942 @xmath943 by using our twisted morse theory . taking a sufficiently generic @xmath944 we set @xmath945 .
then by proposition [ h - dim-1 ] the real - valued function @xmath946 has only @xmath900 non - degenerate ( morse ) critical points in @xmath947 .
moreover we can easily verify that the index of such a critical point is always @xmath892 .
for @xmath948 we define an open subset @xmath949 of @xmath950 by @xmath951 then we have the following result . let @xmath955 be a smooth subdivision of the dual fan of @xmath896 and @xmath956 the ( smooth ) toric variety associated to it .
then the divisor at infinity @xmath957 is normal crossing . by the non - degeneracy of @xmath745 the divisor
@xmath958 is also normal crossing in a neighborhood of @xmath164 .
moreover by the condition @xmath959 the neighborhood @xmath960 of @xmath961 retracts to @xmath962 as @xmath963 . since for generic @xmath964 the local system @xmath756 has a non - trivial monodromy around each irreducible component of @xmath164 , the assertion follows . [ tod ] for generic @xmath952 we have @xmath966 and there exists a basis of @xmath967 indexed by the @xmath900 non - degenerate ( morse ) critical points of the ( possibly multi - valued ) function @xmath945 in @xmath968 . note that theorem [ tod ] partially solves the famous problem in the paper gelfand - kapranov - zelevinsky @xcite of constructing a basis of the twisted homology group in their integral representation of @xmath0-hypergeometric functions .
indeed theorem [ tod ] holds even if we replace the local system @xmath756 with the one @xmath969 on @xmath970 .
now we return to our original situation in sections [ sec:4 ] and [ sec:5 ] .
we fix a point @xmath396 and define @xmath833 etc . in the real oriented blow - up @xmath80 of @xmath834 as in the proof of theorem [ main ] . for the local system @xmath971 on @xmath301
we shall construct a basis of the rapid decay homology group @xmath972 . as in section [ sec:6 ] , for an open subset @xmath360 of @xmath301 and @xmath126
we set @xmath837 for short .
recall that @xmath842 are the rays in the smooth fan @xmath434 which correspond to the relevant divisors @xmath840 in @xmath841 . by the primitive vector @xmath973 on @xmath443 the order @xmath446 of the pole of @xmath846 along @xmath161
is explicitly described by @xmath847 by the non - degeneracy of @xmath19 the complex hypersurface @xmath855 intersects each relevant divisor @xmath161 transversally .
let @xmath974 be the @xmath16-orbit in @xmath435 which corresponds to @xmath443 and denote by the same letter @xmath867 its strict transform in the blow - up @xmath868 .
recall also that for @xmath447 the euler characteristic of the hypersurface @xmath975 in @xmath867 is equal to @xmath976 .
let @xmath604 be the coordinates on an affine chart @xmath977 of @xmath435 containing @xmath867 such that @xmath978 and @xmath979 .
define a local system @xmath980 on @xmath872 by @xmath981 .
if @xmath985 ( @xmath986 ) , the assertion follows immediately from theorem [ tod ] . if @xmath987 ( @xmath988 ) , we have @xmath989 for an open subset @xmath360 of @xmath990 .
hence for generic @xmath991 there exists an isomorphism @xmath992 for any @xmath126 .
[ tak ] for each generic parameter vector @xmath67 and @xmath447 there exists a sufficiently small neighborhood @xmath871 of @xmath872 in @xmath81 such that for its open subset @xmath873 we have @xmath993 and can construct a basis @xmath994 ( @xmath876 , @xmath865 ) of @xmath995 by the twisted morse theory .
let @xmath996 be the function on @xmath997 defined by @xmath998 for a sufficiently generic @xmath999 .
then by proposition [ h - dim-1 ] the function @xmath1000 has only @xmath1001 non - degenerate ( morse ) critical points in @xmath997 . by the product decomposition
@xmath1002 we consider @xmath1000 also as a function on the open subset @xmath1003 of @xmath1004 .
for @xmath948 we set @xmath1005 then it follows from lemma [ 2-dim-1 ] ( i ) that by shrinking @xmath871 and taking large enough @xmath1006 we obtain isomorphisms @xmath1007 in the same way as the proof of proposition [ van ] , for generic @xmath67 and sufficiently small @xmath758 we can show that @xmath1008 @xmath943 . let @xmath1009 be the critical values of @xmath996 .
we may assume that @xmath1010 . for @xmath649 let @xmath1011 be the critical points of @xmath1000 such that @xmath1012 .
then for sufficiently small @xmath653 there exists a long exact sequence @xmath1013 where @xmath1014 is the stable manifold of the gradient flow of @xmath1000 passing through its critical point @xmath1015 and @xmath1016 is the punctured disk @xmath1017 .
moreover we assume here that @xmath1018 are homeomorphic to the @xmath476-dimensional disk . then by lemma [ ecl ]
the dimension of the vector space @xmath1019 is @xmath1020 . as a result
we obtain @xmath1021 ( see ) . since for generic @xmath67 we have the vanishings of the usual twisted homology groups @xmath888 and @xmath1022 etc .
, by mayer - vietoris exact sequences for relative twisted homology groups we obtain also isomorphisms @xmath1023 therefore by we get @xmath993 this implies that for any @xmath649 and generic @xmath67 there exists a short exact sequence @xmath1024 moreover by lemma [ ecl ] we can construct a natural basis of the vector space @xmath1025 .
lifting these bases to @xmath1026 with the help of the above short exact sequences we obtain the one @xmath994 ( @xmath876 , @xmath865 ) of @xmath1027 .
[ h - dim ] for generic nonresonant parameter vectors @xmath67 the natural morphisms @xmath1028 are injective and induce an isomorphism @xmath1029 in particular , the cycles @xmath1030 ( @xmath447 , @xmath876 , @xmath865 ) form a basis of the vector space @xmath1031 over @xmath96 . let @xmath1032 be the set of nonresonant parameters and @xmath1033 its subset consisting of parameters for which theorem [ h - dim ] holds .
it is clear that @xmath1034 is open dense in @xmath1035 . for a point @xmath1036 assume that there exists a point @xmath1037 close to @xmath1038 such that the @xmath585 linearly independent rapid decay @xmath8-cycles constructed at @xmath1039
can be continuously extended to @xmath1038 .
then we can show that the resulting ones at @xmath1038 are also linearly independent as follows .
let @xmath1040 be the local system of rank @xmath585 on @xmath1035 defined by the @xmath8-th rapid decay homology groups .
then the above families of rapid decay @xmath8-cycles define sections @xmath1041 of @xmath1040 on a connected open neighborhood @xmath215 of @xmath1039 such that @xmath1042 .
we thus obtain a homomorphism of sheaves @xmath1043 on @xmath215 whose stalk at @xmath1044 is an isomorphism . by sheaf theory , it is an isomorphism over the whole @xmath215 .
this implies that the @xmath585 rapid decay @xmath8-cycles at @xmath1038 are also linearly independent .
|
we study confluent @xmath0-hypergeometric functions introduced by adolphson @xcite . in particular
, we give their integral representations by using rapid decay homology cycles of hien @xcite and @xcite .
the method of toric compactifications introduced in @xcite and @xcite will be used to prove our main theorem .
moreover we apply it to obtain a formula for the asymptotic expansions at infinity of confluent @xmath0-hypergeometric functions .
|
galaxy clusters are ideal probes for studies of large scale structures .
they have typical masses of order @xmath0 ; within a radius of a few mpc they contain hundreds of galaxies orbiting in a gravitational potential well due primarily to dark matter ( dm ) .
they are also filled with icm : hot ionized gas , typically at a temperature of 1 - 15 kev , that can be observed via its bremsstrahlung emission in the soft x - ray band , and via the sunyaev - zeldovich ( sz ) effect in the millimetric band . several observational evidences
demonstrate that the picture of an icm in hydrostatic equilibrium in the dm potential well of the cluster is an over - simplification : above all , scaling relations between various cluster properties , like mass , temperature and x - ray luminosity , differ from the self - similar predictions ( see , e.g. , the recent reviews by * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and the references therein ) .
in order to explain the icm properties , many models have been developed so far , which consider the effect of radiative cooling and of non - gravitational heating ( mainly feedback from supernovae and agns ) .
as the complexity in the physical description of the icm increased , hydrodynamic simulations have also improved to include such processes , and have become invaluable tools to compare theoretical predictions with current data and to forecast the performance of future experiments . on the observational side ,
important improvements are expected for the detection of the sz effect in clusters thanks to an upcoming new generation of suitable instruments , like ami , spt , act , sza , amiba , apex , and planck . in order to correctly extract and utilize the information contained in these data
it will be mandatory to keep under control both the dependence of the sz signal on the icm physics , and the possible biases induced on the intrinsic signal by the details of the observational process . in the present work
we investigate how different choices for the physical modelling of the icm affect the sz radial profiles and the scaling relations between the sz flux and other intrinsic properties of clusters . to this end
, we use a set of high - resolution hydrodynamical simulations , where clusters were re - simulated including different physical processes , from radiative cooling , to star formation , energy feedback from supernovae , and thermal conduction . after this analysis
, we adopt the instrumental characteristics of the ami interferometer and simulate the observational process for a few clusters in our sample : this enable us to discuss possible biases introduced in our results by a realistic observation process .
the present paper is organized as follows .
after briefly introducing in sect .
[ sec : uno ] the sz effect and the main related quantities , in sect .
[ sec : due ] we describe the general characteristics of the hydrodynamic simulations used in our study . in sect .
[ sec : tre ] we obtain fitting formulas for the cluster sz profiles , while in sect .
[ sec : quattro ] we discuss the effects of various physical processes on the scaling relations between some global properties commonly used to describe galaxy clusters . in sect .
[ sec : cinque ] we perform simulated observations assuming the characteristics of the ami instrument , and discuss the robustness of our results . finally in sect .
[ sec : conclu ] we draw our conclusions .
the thermal sz effect @xcite is a distorsion in the cosmic microwave background ( cmb ) spectrum due to inverse compton scattering of cmb photons by hot ionized gas particles ( for recent reviews see , e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
its intensity is given by the comptonization parameter @xmath1 , defined as : @xmath2 where @xmath3 is the boltzmann constant , @xmath4 is the thomson cross section , @xmath5 is the speed of light , and @xmath6 , @xmath7 and @xmath8 are the electronic number density , temperature and rest mass , respectively . as evident from eq.([eq : y ] ) , the @xmath1-parameter is directly proportional to the cluster pressure @xmath9 integrated along the line of sight .
the change in the cmb intensity at the frequency @xmath10 corresponding to a certain value of the comptonization parameter is : @xmath11 where @xmath12 is the planck constant .
the frequency dependence is given by the spectral shape @xmath13 : @xmath14 where @xmath15 . in galaxy clusters the comptonization parameter @xmath1 typically reaches values of order @xmath16 on arc - minute scales , corresponding to fluctuations in the cmb temperature of few @xmath17 .
as a consequence , observing galaxy clusters through the sz effect requires extremely good sensitivity , in order to obtain sufficiently high signal - to - noise ratios . the required sensitivity and sky coverage will be achieved in the near future , thanks to new dedicated instruments .
nevertheless , the resolution of these instruments will be not enough to detect the wealth of details which can be seen in numerical hydrodynamical simulations : substructures , shocks , etc .
for this reason , in order to be meaningful and helpful , all analyses of the sz effect based on simulations must deal with the actual instrumental capabilities : this is one of the main goals of this paper , as discussed below .
the hydrodynamic simulations used in the present work were carried out using gadget-2 @xcite , a new version of the parallel tree+sph ( smoothed particle hydrodynamics ) code gadget @xcite .
gadget-2 includes an entropy - conserving formulation of sph @xcite , radiative cooling , heating by an uv background , and a treatment of star formation and feedback from galactic winds powered by supernova explosions .
the latter is based on a subresolution model for the multi - phase structure of the interstellar medium , as described in @xcite .
the sph implementation of thermal conduction in gadget-2 , which is both stable and manifestly conserves thermal energy - even when individual and adaptive time - steps are used - was described in @xcite .
this implementation assumes an isotropic effective conductivity parameterized as a fixed fraction of the spitzer rate .
it also accounts for saturation , which can become relevant for low - density gas .
.different physical processes included in our re - simulations [ cols="^,^",options="header " , ] . for comparison ,
the thin lines refer to the corresponding mean logarithmic sz profiles directly extracted from the hydrodynamic simulations .
only the hm sample at @xmath18 is considered here.,scaledwidth=35.0% ]
we used a sample of simulated galaxy clusters with masses ranging from @xmath19 to about @xmath20 , re - simulated using different sets of physical processes ( radiative cooling , star formation , feedback and thermal conduction ) to investigate the dependence of the sz emission on the icm physics .
starting from the 3d pressure information , we obtained for each physics fitting relations to the sz radial profiles .
we compared the scaling relations linking cluster central and total sz flux , and cluster x - ray luminosity and ( mass - weighted ) temperature , with those predicted by the self - similar model .
we then performed mock observations of the most massive systems in our sample , in order to check whether our findings survive the observational procedure .
we summarize our main results below .
* the shape of the sz cluster profile strongly depends on the cluster mass , as a result of the fact that our high mass clusters , at variance with the low mass ones , are all isolated and virialized .
in particular , while the 3d ( logarithmic ) pressure profiles of low - mass clusters ( lm sample ) can be described by a function which is nearly flat for @xmath21 and steepening for @xmath22 , the high - mass clusters ( hm sample ) exhibit an additional steepening of the profiles at @xmath23 . *
the details of the icm physics have little impact on the sz profiles ; the largest differences are found near the central regions of low - mass systems . *
the scaling relations between cluster sz emission and cluster x - ray luminosity and temperature evolve with redshift according to the self - similar model out to @xmath24 ; this result is consistent with previous works @xcite . *
the slopes of the scaling relations for all physics are consistent with the predictions of the self - similar model , even if small discrepancies can still be found : in particular , our empirical relations @xmath25 and @xmath26 are generally slightly steeper than the self - similar ones , while the @xmath27 relation is generally slightly shallower .
* the icm physics has little effect on the slopes of the scaling relations , in agreement with previous studies @xcite .
* the main effect introduced in the scaling relations by radiative cooling is a lower normalization .
if we compare our simulations including cooling with the non - radiative runs with standard viscosity ( _ ovisc _ ) , we find a suppression of @xmath28 per cent in the central intensity @xmath29 and of @xmath30 per cent in the integrated flux @xmath31 . if we compare the cooling plus star - formation runs with the non - radiative ones adopting modified artificial viscosity ( _ lvisc _ ) , these reductions are of the order of @xmath32 per cent . a similar normalization lowering for the @xmath27 relation was also found by @xcite .
* our mock observations suggest that arcminute resolution data from the current generation of sze imaging experiments will not be a sensitive probe of the different physics that is ongoing in the intracluster medium . at arcminute scale angular resolution
these effects tend to only have observable consequences on the total cluster flux , which leads to a degeneracy between physics and cluster mass that will be difficult to break * the integrated flux @xmath31 is found to be stable against the observational process , and thus should be preferred to @xmath29 , which instead exhibits large variations after performing the mock observations .
we thank e. rasia , m. massardi , p. mazzotta , s. borgani and r. kneissl for useful discussions and the anonymous referee for her / his comments which allowed to improve the presentation of our results .
we aknowledge k. grainge and the ami collaboration for providing all the necessary information about the ami interferometer .
the simulations have been performed using the ibm - sp4/5 machine at the `` consorzio interuniversitario del nord - est per il calcolo elettronico '' ( cineca , bologna ) , with cpu time assigned thanks to the inaf
cineca grant , the ibm - sp3 machine at the italian centre of excellence `` science and applications of advanced computational paradigms '' ( padova ) and the ibm - sp4 machine at the `` rechenzentrum der max - planck - gesellschaft '' at the `` max - planck - institut fr plasmaphysik '' with cpu time assigned to the `` max - planck - institut fr astrophysik '' .
kd acknowledges partial support by a marie curie fellowship of the european community program `` human potential '' under contract number mcfi-2001 - 01227 .
m. e. , 2002 , in chen l .- w .
, ma c .- p . , ng k .- w . ,
, eds , asp conf .
ser . 257 : amiba 2001 : high - z clusters , missing baryons , and cmb polarization sz surveys with the arcminute microkelvin imager .
pp 35+ k. y. , 2002 , in chen l .- w . , ma c .- p .
, ng k .- w .
, eds , asp conf . ser . 257 : amiba 2001 : high - z clusters , missing baryons , and cmb polarization the current state of astronomy in taiwan and amiba .
pp 3+ j. j. , carlstrom j. e. , the sza collaboration 2002 , in chen l .- w . , ma c .- p . , ng k .- w . ,
, eds , asp conf . ser . 257 : amiba 2001 : high - z clusters , missing baryons , and cmb polarization the sz - array : configuration and science prospects .
pp 43+ j. , et al . , 2004 , in zmuidzinas j. , holland w. s. , withington s. , eds , astronomical structures and mechanisms technology . edited by antebi , joseph ; lemke , dietrich .
proceedings of the spie , volume 5498 , pp .
11 - 29 ( 2004 ) .
the south pole telescope .
|
we use high - resolution hydrodynamic re - simulations to investigate the properties of the thermal sunyaev - zeldovich ( sz ) effect from galaxy clusters . we compare results obtained using different physical models for the intracluster medium ( icm ) , and
show how they modify the sz emission in terms of cluster profiles and scaling relations .
we also produce realistic mock observations to verify whether the results from hydrodynamic simulations can be confirmed .
we find that sz profiles depend marginally on the modelled physical processes , while they exhibit a strong dependence on cluster mass .
the central and total sz emission strongly correlate with the cluster x - ray luminosity and temperature .
the logarithmic slopes of these scaling relations differ from the self - similar predictions by less than 0.2 ; the normalization of the relations is lower for simulations including radiative cooling .
the observational test suggests that sz cluster profiles are unlikely to be able to probe the icm physics .
the total sz decrement appears to be an observable much more robust than the central intensity , and we suggest using the former to investigate scaling relations .
[ firstpage ] cosmic microwave background galaxies : clusters : general hydrodynamics methods : numerical
|
low surface brightness galaxies ( lsbgs ) are important populations in galaxy field .
however , their contributions to galaxy population have been underestimated for a long time since they are hard to find owing to their faintness compared with the night sky .
an initial quantitative study was made by freeman ( 1970 ) , who noticed that the central surface brightness of their 28 out of 36 disc galaxies fell within a rather narrow range , @xmath0(b)[email protected] mag arcsec@xmath1 .
this could be caused by selection effects ( disney 1976 ) .
since then , many efforts have been made to search for large number of lsbgs from surveys ( bothun & impey 1997 ; impey & bothun 1997 ; zhong et al .
2008 and references therein ) .
one of the important ones is the apm survey , i.e. , impey et al .
adopted the automated plate measuring ( apm ) mechanism to scan uk schmidt plates and discovered 693 lsbgs which forms the most extensive catalog of lsbgs to that date ( impey et al .
oneil et al .
( 1997 , in texas survey " ) firstly found the red lsbg populations .
monnier ragaigne et al .
( 2003 ) selected a sample of about 3800 lsbgs from the all - sky near - infrared 2mass survey , and then made hi observations for a sub - sample .
the modern digital sky survey , such as the sloan digital sky survey ( sdss ) , certainly provide a wonderful chance for us to find much more lsbgs .
kniazev et al .
( 2004 ) developed a method to search for lsbgs from sdss images and used the apm sample to test their method .
they recover 87 same objects and 42 new lsbgs .
recently , our group ( zhong et al .
2008 ) successfully select a large sample of lsbgs from sdss - dr4 main galaxy sample , which includes 12,282 nearly face - on disk galaxies with low surface brightness @xmath0(b)@xmath3 22 mag arcsec@xmath1 , and another 18,051 high surface brightness galaxies ( hsbgs ) with @xmath0(b)@xmath422 mag arcsec@xmath1 are also selected for comparisons .
this will be a very efficient sample to study the properties of lsbgs , and to check the variation of these properties with surface brightnesses if there exists .
we have studied some properties of this large sample .
zhong et al .
( 2008 ) presented the sample selection criteria , and their magnitudes , surface brightness , scalelengths and colors etc .
the metallicities ( liang et al .
2009 , in preparation ) and stellar populations ( chen et al .
2009 , in preparation ) are also studied from optical spectra , i.e. the emission lines , continua and absorption lines .
in addition , the modern digital sky surveys have been made in wide wavelength ranges , such as galex - uv , sdss - optical , 2mass - nir , iras - ir etc .
these allow us to construct the multi - wavelength seds for our sample galaxies ( gao et al .
2009 , in preparation ) , which are very useful and efficient to study their star formation history . in this paper
, we summarize the interesting results we obtained for this large sample of galaxies , which will be presented in detail in the series work mentioned above .
especially , we try to check the varying trends of their properties following surface brightness . a cosmological model with @xmath5 = 70 km s@xmath6 mpc@xmath6 , @xmath7 = 0.3 and @xmath8 = 0.7 is adopted in this paper .
we select 30,333 nearly face - on disk galaxies from the sdss - dr4 main galaxy sample .
they have been estimated the reliable surface brightness values and minimize the effect of dust extinction inside the galaxies .
the selection criteria are given below . 1 .
@xmath12 @xmath4 0.25 ( @xmath12 indicates the fraction of luminosity contributed by the de vaucouleurs profile relative to exponential profile in the @xmath13-band , and this low value for limit can also minimize the effect of bulge light on the disk galaxies ) ; 2 .
@xmath14 @xmath15 0.75 ( for nearly face - on galaxies , where @xmath16 and @xmath17 are the semi - major and semi - minor axes of the fitted exponential disk respectively ) ; 3 .
@xmath9 @xmath4 -18 ( this excludes the few ( 6% ) dwarf galaxies contained in the sample ) . taking @xmath0(b)=22 mag arcsec@xmath1 as criterion , 12,282 objects are confirmed as lsbgs and other 18,051 objects are confirmed as hsbgs .
the properties of this large sample of lsbgs have been studied carefully by zhong et al .
specially , these lsbgs show clear correlations between @xmath9 and log@xmath10 , and between log@xmath11 and log@xmath10 , which mean that the brighter galaxies tend to have larger scalelengths and the galaxies with larger scalelengths have the benefit to be detected and observed in the more distant universe .
furthermore , both of the lsbgs and hsbgs can be divided into two sub - groups following @xmath0(b ) , which will allow us to study their properties in a sequence of surface brightness .
the criteria for @xmath18 ( in mag arcsec@xmath1 ) are following mcgaugh ( 1996 ) : slsbgs ( 2,815 ) with 22.75 - 24.5 , isbgs ( 9,467 ) with 22.0 - 22.75 , shsbgs ( 10,989 ) with 21.25 - 22.0 , vhsbgs ( 7,062 ) with @xmath421.25 .
fig.1a shows the relations of @xmath9 vs. log@xmath10 for the samples in these four sub - groups .
fig.1b shows that for log@xmath11 vs. log@xmath10 .
all the four sub - groups show clear correlations . however ,
the comparisons among these four show that , at a given disk scalelength , the galaxies having higher @xmath0(b ) are brighter , and could be observed at a farther distance
. there are gradually varying trends following surface brightnesses of the galaxies .
metallicity is an important parameter of galaxy and a good tracer for their evolution history .
the metallicities of this large sample of nearly face - on disk galaxies can be estimated from optical emission lines . to select the target galaxies with good quality spectroscopic observations , we adopt the criteria below ( following liang et al .
2006 and tremonti et al .
2004 ) . 1
. cross correlating between our sample and the sdss - dr4 spectroscopic database .
the fluxes of the emission lines of the galaxies have been measured for [ o ii ] , h@xmath19 , [ o iii ] , h@xmath20 , [ n
ii ] , and the line h@xmath19 , h@xmath20 , and [ n ii]6584 are detected with s / n ratios greater than 5@xmath21 .
then the selected sample in the four sub - groups are 1,364 in slsbgs , 6,055 in isbgs , 9,107 in shsbgs are 6,231 in vhsbgs .
3 . selecting the star - forming galaxies by using the diagnostic diagram of [ n ii]/h@xmath20 vs. [ o iii]/h@xmath19 following kauffmann et al .
( 2003a ) .
now we obtain 1,299 of slsbgs , 5,551 of isbgs , 8,310 of shsbgs , and 5,872 of vhsbgs .
this also means that the agn fractions of these sub - groups are quite small .
the 12+log(o / h ) abundances of these star - forming galaxies are taken from the mpa/ jhu database ( tremonti et al .
the median values in the four sub - groups are 8.77 for slsbgs , 8.94 for isbgs , 9.03 for shsbgs , and 9.06 for vhsbgs .
this means that the galaxies have lower surface brightness generally have lower metallicity .
fig.2a shows the relations of 12+log(o / h ) vs. @xmath18 of the star - forming galaxies in the four sub - groups .
the median values in each of the bins of 0.2 in @xmath18 are also given ( the big squares ) , as well as the three vertical long - dashed lines at @xmath18=21.25 , 22.0 and 22.75 to mark the ranges of @xmath18 for the four sub - groups .
a general varying trend shows that , for the slsbgs , isbgs and shsbgs , the galaxies with lower surface brightness have lower metallicity .
we further obtain the stellar masses of these sample galaxies from the mpa / jhu database ( kauffmann et al .
2003b ) , and present their relations with @xmath18 in fig.2b .
it shows that , for the slsbgs , isbgs and shsbgs , the galaxies with lower surface brightness have smaller stellar mass generally .
these are in agreement with the stellar mass - metallicity relations of the galaxies .
epsf from 18.0 - 24.5 mag arcsec@xmath1 , ( b ) correlation of disk scale length and distances for them : slsbgs ( 24.5 - 22.75 , the solid line ) ; isbgs ( 22.75 - 22.0 , the dot - long dash line ) ; shsbgs ( 22.0 - 21.25 , the long - dash line ) ; vhsbgs ( @xmath421.25 , the short - long dash line ) . , title="fig:",width=207,height=170 ] from 18.0 - 24.5 mag arcsec@xmath1 , ( b ) correlation of disk scale length and distances for them : slsbgs ( 24.5 - 22.75 , the solid line ) ; isbgs ( 22.75 - 22.0 , the dot - long dash line ) ; shsbgs ( 22.0 - 21.25 , the long - dash line ) ; vhsbgs ( @xmath421.25 , the short - long dash line ) . , title="fig:",width=207,height=170 ] for the star - forming galaxies , ( b ) the relations of stellar mass vs. @xmath18 of them . the median values in each of the bins of 0.2 in @xmath18
are also given as the big squares .
the three vertical long - dashed lines at 22.75 , 22.0 and 21.25 mark the ranges of @xmath18 for the four sub - groups of slsbgs , isbgs , shsbgs , vhsbgs.,title="fig:",width=207,height=170 ] for the star - forming galaxies , ( b ) the relations of stellar mass vs. @xmath18 of them .
the median values in each of the bins of 0.2 in @xmath18 are also given as the big squares .
the three vertical long - dashed lines at 22.75 , 22.0 and 21.25 mark the ranges of @xmath18 for the four sub - groups of slsbgs , isbgs , shsbgs , vhsbgs.,title="fig:",width=207,height=170 ]
we study the stellar populations of this sample of star - forming galaxies by using the software starlight ( cid fernandes et al . 2005 ; chen et al .
2009 ) to fit the optical spectral absorptions and continua .
the template for spectral synthesis analysis are the simple stellar populations from bruzual & charlot ( 2003 ) .
we combined the spectra of all the objects to be one spectrum in each of the sub - samples , then we synthesize these combined spectra .
the results show that the importance of young populations increases from slsbgs to isbgs , shsbgs , and to vhsbgs ( more details in chen et al .
2009 , in preparation ) .
since the multiwavelength modern digital sky surveys have observed and released a very large sample of objects to public , such as galex - uv , sdss - optical , 2mass - nir and iras - ir ( the spitzer / swire survey also release their ir data , but in a much smaller sky coverage ) .
therefore , we cross correlate our such large sample of nearly face - on disk galaxies selected from optical with the observations at uv , nir and ir , and then construct their multiwavelength seds .
we extend the sdss - dr4 to the dr7 main galaxy sample here , which expands to be 21,666 lsbgs and 30,898 hsbgs by following the same criteria given in sect.2 .
finally we obtain about 280 lsbgs and 300 hsbgs with reliable observations from uv to ir . if without considering ir , there are about 2100 lsbgs and 7300 hsbgs having multiwavelength observations from uv to nir
. these will be good samples to study the multiwavelength seds and stellar formation history of galaxies .
one of the useful model to analyze their seds is grasil ( silva et al . 1998 ) .
this model could produce the multiwavelength seds of galaxies , such as m51 ( nearly face - on disk ) , m82 ( starburst ) and ngc 6090 ( strongly interacting galaxy ) etc .
our sample could be compared with these seds and hopefully they will be similar to those of m51 and/or m82 . in summary , we have selected a large sample of lsbgs from sdss , which could provide interesting and important results for understanding the properties of this kind of galaxies .
also it must be very useful to better understand their contribution to the local galaxy population .
their effects on the number density , light density and light functions of galaxies in the local universe will be further studied .
|
a large sample of low surface brightness ( lsb ) disk galaxies is selected from sdss with b - band central surface brightness @xmath0(b ) from 22 to 24.5 mag arcsec@xmath1 .
some of their properties are studied , such as magnitudes , surface brightness , scalelengths , colors , metallicities , stellar populations , stellar masses and multiwavelength seds from uv to ir etc .
these properties of lsb galaxies have been compared with those of the galaxies with higher surface brightnesses .
then we check the variations of these properties following surface brightness .
|
much attention has been directed by theoretical / computational chemists towards developing reliable and accurate means for solving dynamical quantum mechanics problems i.e . , for obtaining solutions to the time - dependent equation for molecular systems .
insofar as `` exact '' quantum methods are concerned , two traditional approaches have been used : ( 1 ) representation of the system hamiltonian in a finite , direct - product basis set ; ( 2 ) discretization of the wavefunction onto a rectilinear grid of lattice points over the relevant region of configuration space . both approaches
, however , suffer from the drawback that the computational effort scales exponentially with system dimensionality.@xcite recently , a number of promising new methods have emerged with the potential to alleviate the exponential scaling problem once and for all .
these include various basis set optimization methods,@xcite and build - and - prune methods,@xcite such as those based on wavelet techniques.@xcite on the other hand , a completely different approach to the exponential scaling problem is to use basis sets or grid points , that themselves evolve over time .
the idea is that at any given point in time , one need sample a much smaller hilbert subspace , or configuration space region , than would be required at all times thus substantially reducing the size of the calculation . for basis set calculations , much progress along these lines has been achieved by the multi - configurational time - dependent hartree ( mctdh ) method , developed by meyer , manthe and co - workers.@xcite more recently , time - evolving grid , or `` quantum trajectory '' methods@xcite ( qtms ) have also been developed , and for certain types of systems , successfully applied at quite high dimensionalities.@xcite qtms are based on the hydrodynamical picture of quantum mechanics , developed over half a century ago by bohm@xcite and takabayasi,@xcite who built on the earlier work of madelung@xcite and van vleck.@xcite qtms are inherently appealing for a number of reasons .
first , they offer an intuitive , classical - like understanding of the underlying dynamics , which is difficult - to - impossible to extract from more traditional fixed grid / basis methods . in effect , quantum trajectories are like ordinary classical trajectories , except that they evolve under a modified potential @xmath4 , where @xmath5 is the wavefunction - dependent `` quantum potential '' correction .
second , qtms hold the promise of delivering exact quantum mechanical results without exponential scaling in computational effort .
third , they provide a pedagogical understanding of entirely quantum mechanical effects such as tunneling@xcite and interference.@xcite they have already been used to solve a variety of different types of problems , including barrier transmission,@xcite non - adiabatic dynamics,@xcite and mode relaxation.@xcite several intriguing phase space generalizations have also emerged,@xcite of particular relevance for dissipative systems.@xcite despite this success , qtms suffer from a significant numerical drawback , which , to date , precludes a completely robust application of these methods .
namely : qtms are numerically unstable in the vicinity of amplitude nodes .
this `` node problem '' manifests in several different ways:@xcite ( 1 ) infinite forces , giving rise to kinky , erratic trajectories ; ( 2 ) compression / inflation of trajectories near wavefunction local extrema / nodes , leading to ; ( 3 ) insufficient sampling for accurate derivative evaluations .
nodes are usefully divided into two categories,@xcite depending on whether @xmath5 is formally well - behaved ( `` type one '' nodes ) or singular ( `` type two '' nodes ) . for stationary state solutions to the equation , for instance , all nodes are type one nodes . in principle ,
type one nodes are `` gentler '' than type two nodes ; however , from a numerical standpoint , even type one nodes will give rise to the problems listed above , because the slightest numerical error in the evaluation of @xmath5 is sufficient to cause instability .
in the best case , the node problem simply results in substantially more trajectories and time steps than the corresponding classical calculation ; in the worst case , the qtm calculation may fail altogether , beyond a certain point in time .
several numerical methods , both `` exact '' and approximate , are currently being developed to deal with this important problem .
the latter category includes the artificial viscosity@xcite and linearized quantum force methods,@xcite both of which have proven to be very stable . while such approximate methods may not capture the hydrodynamic fields with complete accuracy in nodal regions , they do allow for continued evolution and long - time solutions , often unattainable via use of a traditional qtm .
the `` exact '' methods include the adaptive hybrid methods,@xcite and the complex amplitude method.@xcite in the adaptive hybrid methods , for which hydrodynamic trajectories are evolved everywhere except for in nodal regions , where the time - dependent schrodinger equation is solved instead to avoid node problems .
although they have been applied successfully for some problems , these methods are difficult to implement numerically , since not only must the hydrodynamic fields be somehow monitored for forming singularities , but there must also be an accurate means for interfacing and coupling the two completely different equations of motion . the complex amplitude method is cleaner to implement , but is only exact for linear and quadratic hamiltonians .
in a recent paper,@xcite hereinafter referred to as `` paper i , '' one of the authors ( poirier ) introduced a new strategy for dealing with the node problem , based on a bipolar decomposition of the wavefunction .
the idea is to partition the wavefunction into two ( or in principle , more ) component functions , i.e. @xmath6 .
one then applies qtm propagation separately to @xmath2 and @xmath3 , which can be linearly superposed to generate @xmath1 itself at any desired later time .
in essence , this works because the equation itself is linear , but the equivalent bohmian mechanical , or quantum hamilton s equations of motion ( qhem ) are not.@xcite in principle , therefore , one may improve the numerical performance of qtm calculations simply by judiciously dividing up the initial wavepacket into pieces .
although bipolar decompositions have been around for quite some time,@xcite their use as a tool for circumventing the node problem for qtm calculations is quite recent .
two promising new exact methods that seek to accomplish this are the so - called `` counter - propagating wave '' method ( cpwm),@xcite and the `` covering function '' method ( cfm).@xcite in the cpwm , the bipolar decomposition is chosen to correspond to the semiclassical wkb approximation,@xcite for which all of the hydrodynamic field functions are smooth and classical - like , and the component wavefunctions are node - free . interference is achieved naturally , via the superposition of left- and right - traveling ( i.e. positive- and negative - momentum ) waves . for one - dimensional ( 1d ) stationary bound states , it can be shown that the resultant bipolar quantum potential @xmath7 becomes arbitrarily small in the large action limit , even though the number of nodes becomes arbitrarily large .
( note : in accord with the convention established in ref .
, upper / lower case will be used to denote the unipolar / bipolar field quantities ) . in the cfm ,
the idea is to superpose some well - behaved large - amplitude wave , with the actual ill - behaved ( nodal or wildly oscillatory ) wave , so as to `` dilute '' the undesirable numerical ramifications of the latter .
this paper is the second in a series designed to explore the cpwm approach , introduced in paper i. as discussed there in greater detail , there are many motivations for this approach , but the primary one is to reconcile the semiclassical and bohmian theories , in a manner that preserves the best features of both , and also satisfies the correspondence principle . for our purposes
, this means that the lagrangian manifolds ( lms ) for the two theories should become identical in the large action limit ( sec .
[ scattering ] ) .
as described above , a key benefit of the cpwm decomposition is an elegant treatment of interference , the chief source of nodes and `` quasi - nodes''@xcite ( i.e. rapid oscillations ) in quantum mechanical systems .
an interesting perspective on the role of interference in semiclassical and bohmian contexts is to be found in a recent article by zhao and makri.@xcite whereas paper i focused on stationary bound states for 1d systems , the present paper ( paper ii ) and the next in the series ( paper iii)@xcite concern themselves with stationary scattering states .
the cpwm decomposition of paper i is uniquely specified for any arbitrary 1d state bound or scattering and in the bound case , always satisfies the correspondence principle .
however , the non-@xmath8 nature of the scattering states is such that the paper i decomposition generally does _ not _ satisfy the correspondence principle in this case .
simply put , the quantum trajectories and lms exhibit oscillatory behavior in at least one asymptotic region ( thereby manifesting reflection ) , whereas the semiclassical lms do not .
this is not a limitation of the cpwm , but is rather due to the fundamental failure of the basic wkb approximation to predict any reflection whatsoever for above - barrier energies , as has been previously well established.@xcite in semiclassical theory , a modification must therefore be made to the basic wkb approximation , in order to obtain meaningful scattering quantities . as discussed in sec .
[ scattering ] and in paper iii , our approach will be to apply a similar modification to the exact quantum decomposition ( actually , a _ reverse _ modification ) such that the correspondence principle remains satisfied , and the two theories thus reconciled , even for scattering systems .
it will be shown the modified cpwm gives rise to bipolar bohmian lms that are _ identical _ to the semiclassical lms , regardless of whether or not the action is large .
put another way , this means that the bipolar quantum potentials @xmath9 effectively _ vanish _ , so that the resultant quantum trajectory evolution is _
completely classical_. moreover , the resultant component wavefunctions , @xmath10 and @xmath11 , correspond asymptotically to the familiar `` incident , '' `` transmitted , '' and `` reflected '' waves of traditional scattering theory .
thus , the modified cpwm implementation of the bipolar bohmian approach provides a natural generalization of these conceptually fundamental entities _ throughout all of configuration space _
, not just in the asymptotic regions , as is the case in conventional quantum scattering theory .
the above conclusions will be demonstrated for both discontinuous and continuous potential systems , in papers ii and iii , respectively
. discontinuous potentials e.g .
the hard wall , the step potential , and the square barrier / well serve as a useful benchmark for the modified cpwm approach , because the scattering component waves ( e.g. `` incident wave , '' etc . ) in this case _ are _ well - defined throughout all of configuration space , according to a conventional scattering treatment .
although this is no longer true for continuous potentials , the foundation laid here in paper ii can be extended to the continuous ( and also time - dependent ) case as well , as described in paper iii .
additional motivation for the development of a scattering version of the cpwm , vis - a - vis the relevance for chemical physics applications , is provided in paper iii .
additional motivation for the consideration of discontinuous potentials is provided in sec .
[ scattering ] of the present paper .
according to the bohmian formulation,@xcite the qhem are derived via substitution of the 1d ( unipolar ) wavefunction ansatz , ( x , t ) = r(x , t ) e^i s(x , t)/ into the time - dependent equation . for the 1d hamiltonian ,
h = -^2 2 m ^2 x^2 + v(x ) , [ hameqn ] this results in the coupled pair of nonlinear partial differential equations , where @xmath12 is the mass , @xmath13 is the system potential , and primes denote spatial partial differentiation .
the first of the two equations above is the quantum hamilton - jacobi equation ( qhje ) , whose last term is equal to @xmath14 , i.e. comprises the quantum potential correction .
the second equation is a continuity equation .
when combined with the quantum trajectory evolution equations , i.e. the continuity equation ensures that the probability [ i.e. density , @xmath15 , times volume element ] carried by individual quantum trajectories is conserved over the course of their time evolution . in paper
i , we derived a unique bipolar decomposition , ( x ) = _ + ( x ) + _ -(x ) , [ bipolardecomp ] for stationary eigenstates @xmath16 of 1d hamiltonians of the form , such that : 1 .
@xmath17 are themselves ( non-@xmath8 ) solutions to the equation , with the same eigenvalue , @xmath18 , as @xmath16 itself .
the invariant flux values , @xmath19 , of the two solutions , @xmath17 , equal those of the two semiclassical ( wkb ) solutions .
the median of the enclosed action , @xmath20 , equals that of the semiclassical solutions .
there are other important properties of the @xmath17,@xcite as discussed in sec .
[ intro ] , and in ref . .
nevertheless , the above three conditions are sufficient to uniquely specify the decomposition . in the special case of
bound ( i.e. @xmath8 ) stationary states , the real - valuedness of @xmath16 implies that the @xmath17 are complex conjugates of each other .
it is natural to ask to what extent the above analysis may be generalized for scattering potentials . certainly , @xmath16 itself is no longer @xmath8 , nor even real - valued , and there are generally two linearly independent solutions of interest for each @xmath18 , instead of just one .
condition ( 1 ) above poses no difficulty for @xmath17 , as these component functions are non-@xmath8 and complex - valued , even in the bound eigenstate case . in principle , condition ( 2 ) is not difficult either ; although the flux value depends on the normalization of @xmath1 itself , which is not @xmath8 , certain well - established normalization conventions for scattering states exist , that can be applied equally well to semiclassical and exact quantum solutions .
there is no action median _ per se _ for scattering states , as the action enclosed within the @xmath21 and @xmath22 phase space lagrangian manifolds@xcite ( lms ) is infinite ; however , the scattering analog of condition ( 3 ) is related to the asymptotic boundary conditions , and it is here that one encounters difficulty .
moreover , an additional concern is raised by the doubly - degenerate nature of the continuum eigenstates , namely : should each scattering @xmath16 have its _ own _
@xmath17 decomposition , or should there be a single @xmath17 pair , from which all degenerate @xmath16 s may be constructed via arbitrary linear superposition ? to resolve these issues , we will adopt the same general strategy used in paper i , i.e. we will resort to semiclassical theory as our guide , wherever possible .
we will also exploit certain special features of the scattering problem not found in generic bound state systems , such as the asymptotic potential condition @xmath23 as @xmath24 ( where primes denote spatial differentiation ) , and its usual implications for scattering theory and applications.@xcite the basic wkb solutions are given by ^(x ) = r_(x ) e^i s_(x)/ , [ scsoln ] where r_(x ) = s_(x ) = [ scrs ] the corresponding positive and negative momentum functions , specifying the semiclassical lms , are given by @xmath25
. equations ( [ scsoln ] ) and ( [ scrs ] ) apply to both bound and scattering cases ; note that for both , @xmath26 are complex conjugates of each other .
the asymptotic potential condition ensures that these approach exact quantum plane waves asymptotically , with the usual scattering interpretations , i.e. @xmath21 in the @xmath27 asymptotic region is the incoming wave from the left ( usually taken to be the incident wave ) , @xmath21 as @xmath28 is the outgoing wave from the left ( the usual transmitted wave ) , etc . insofar as determining the corresponding exact quantum solutions @xmath17 , the procedure described in paper i is still appropriate for bound and semi - bound ( i.e. on one side only ) states , in that the results satisfy the correspondence principle globally , as desired ( for semi - bound examples , consult the appendix ) . for true scattering states
, however , this procedure fails , in the sense that if @xmath21 is chosen to match the normalization and flux of @xmath29 in the @xmath28 asymptote , then it will necessarily approach a nontrivial linear superposition of @xmath29 and @xmath30 in the @xmath31 asymptote , and vice - versa .
there is therefore an ambiguity as to how the corresponding quantum @xmath17 s should be defined , i.e. which asymptotic region should be used to effect the correspondence . more significantly
though , _ either _ choice will result in component functions @xmath17 with substantial interference in one of the two asymptotic regions .
this is due to partial reflection of the exact quantum scattering states , which is not predicted by the basic wkb approximation .
thus , in the large action limit , the exact quantum solutions manifest large - magnitude quantum potentials , @xmath32 , and rapidly oscillating field functions @xmath32 , @xmath33 , and @xmath34exactly the undesirable behavior that the cpwm was introduced to avoid whereas the corresponding basic wkb functions are smooth , and asymptotically uniform .
the lack of any partial reflection is a well - understood shortcoming of the wkb approximation@xcite i.e .
, the basic @xmath26 components , though elegantly constructed from smooth classical functions @xmath35 and @xmath36 , do not in and of themselves correspond to any actual quantum scattering solutions @xmath16 . in light of the bipolar decomposition ideas introduced in paper i , however , our perspective is the reverse one : for any _ actual _ quantum @xmath16 , can one determine an decomposition such that the resultant @xmath17 resemble their well - behaved semiclassical counterparts , and is such a decomposition unique ?
among other properties,@xcite the @xmath17 lm s should become identical to the semiclassical lm s in the large action limit , so as to satisfy the correspondence principle .
based on the considerations of the previous paragraph it is clear that the paper i decomposition does not achieve this goal , when applied to stationary scattering states .
we defer a full accounting of these issues in the context of completely arbitrary continuous potentials @xmath13to paper iii , wherein it will be demonstrated how to compute exact quantum reflection and transmission probabilities ( and stationary scattering states ) using only classical trajectories , and without the need for explicit numerical differentiation of the wavefunction . in the present paper
, we lay the foundation for paper iii , by focusing attention onto two key aspects whose development comprises an essential prerequisite .
first , as the paper iii approach treats @xmath13 as a sequence of steps,@xcite the present paper ii will focus exclusively on the step potential and related discontinuous potential systems , for which @xmath37 in between successive steps .
discontinuous potentials are important for chemical physics , because they model steep repulsive wells , and are used in statistical theories of liquids .
moreover , they hold a special significance for qtm methods , for which they serve as a `` worst - case scenario '' benchmark . indeed , conventional qtm techniques _
always _ fail when applied to discontinuous potentials . to date
, the only such calculations that have been performed@xcite have computed the quantum potential from a completely separate time - dependent fixed - grid calculation ( the `` analytical approach'')@xcite rather than directly from the quantum trajectories themselves .
even if one _ could _ propagate trajectories for discontinous systems using a traditional qtm , the trajectories that would be generated would be very kinky and erratic,@xcite and a great many time trajectories and time steps would thus be required .
second , since the new @xmath17 do _ not _ satisfy condition ( 1 ) , unlike the paper i cpwm decomposition , the time evolution of these two component functions is clearly not that of the time - dependent equation . moreover , since the @xmath38 are constant over time [ because @xmath39 itself is stationary , and is presumed unique ] , _ the two @xmath40 time evolutions must be coupled together_. it is essential that the nature of this coupling be completely understood , in order that the present approach may be generalized to non - stationary state situations e.g . to wavepacket scattering , as will be discussed in future publications .
the ramifications for qtms are equally important .
accordingly , the present paper focuses on the qtm propagation of the wavefunction and its bipolar components with a keen eye towards generality and physical interpretation even though the states involved are stationary .
this approach leads to a pedagogically useful reinterpretation of `` incident , '' `` transmitted , '' and `` reflected '' waves very reminiscent of ray optics in electromagnetic theory which is applicable much more generally than traditional usage might suggest .
the necessary theory will be developed over the course of a consideration of various model application systems of increasing complexity .
let us first consider the simplest case imaginable , the free particle system , @xmath41 . in this case , the exact solutions @xmath42 clearly satisfy the conditions of sec .
[ bipolar ] , and the bipolar quantum potentials @xmath32 are zero everywhere .
thus , the bipolar decomposition developed for bound states in paper i can be used directly with this continuum system , requiring only the slight modification that arbitrary linear combinations of @xmath43 and @xmath44 are to be allowed , in order to construct arbitrary scattering solutions @xmath16 . for convenience ,
the linear combination coefficients will from here on out be directly incorporated into the amplitude functions , @xmath33 , and phase functions , @xmath45 , so that is still correct . if from all solutions @xmath16 one considers only that which satisfies the usual scattering boundary conditions ( i.e. incident wave incoming from the left ) then the negative momentum wave @xmath46 vanishes , and @xmath47 .
there is zero reflection , and @xmath48 transmission .
put another way , the incident flux , @xmath49 , is equal to the transmitted flux , @xmath50 , where [ both flux values are equal to @xmath51 , as in ] . in the quantum trajectory description ,
flux manifests as probability - transporting trajectories , which move along the lms . for the boundary conditions described above
, there are only positive momentum trajectories , moving uniformly from left to right with momentum @xmath52 .
if a @xmath22 contribution were present , its trajectories would move uniformly in the opposite direction [ @xmath53 . ] since the two components @xmath17 are in this case uncoupled , the positive and negative momentum trajectories would have no interaction with each other .
we next consider the hard wall system : v(x ) = in the @xmath54 region , the two @xmath17 components are exactly the same as in the free particle case , except that the @xmath55 boundary condition imposes the additional constraints , s_-(0 ) = s_+(0 ) + ( 2 ) ; r_-(0 ) = r_+(0 ) .
[ hwconst ] this also results in only _
one _ linearly independent solution instead of two , i.e. @xmath56 , with @xmath57 .
regarding the lms and trajectories , in the @xmath58 region , these are identical to those of sec . [ free ] , e.g. the @xmath21 lm trajectories move uniformly to the right , towards the hard wall at @xmath59 .
it is natural to ask what happens when the @xmath21 lm trajectories actually reach @xmath59 .
there are two reasonable interpretations .
the first is that the trajectories keep moving uniformly into the @xmath60 region of configuration space .
this approach treats the hard wall system as if it were the free particle system , but with the @xmath60 region effectively ignored.@xcite this underscores the fact that unlike @xmath16 itself , the individual @xmath17 components _ per se _ are unconstrained at the origin though the constraint implies a unique correspondence between the two .
this interpretation also makes it clear that for the hard wall system , the paper i decomposition is essentially identical to the present decomposition , as is worked out in detail in the appendix . in the second interpretation ,
the effect of the hard wall at @xmath59 is to cause instantaneous elastic reflection of a @xmath21 lm trajectory momentum , from @xmath61 to @xmath62 .
afterwards , the reflected trajectory propagates uniformly backward , along the @xmath22 lm . in this interpretation
, the trajectories never leave the allowed configuration space , @xmath54 .
however , wavepacket reflection is essentially achieved via trajectory _ hopping _ from one lm to the other not unlike that previously considered , e.g. , in the context of non - adiabatic transitions.@xcite the trajectory hopping interpretation is adopted in the present paper , and in paper iii , but the first interpretation will also be reconsidered in later publications . note that for discontinuous potentials and indeed more generally@xcite one can regard trajectory hopping as the _ source _ of @xmath17 interaction coupling .
for the hard wall case , trajectory hopping only manifests at @xmath59 , the sink of all @xmath21 lm trajectories , and the source of all @xmath22 lm trajectories .
if these trajectories are to be regarded as one and the same via hopping , then a unique field transformation for @xmath63 , @xmath64 , and all spatial derivatives , must be specified .
fortunately , the unique correspondence between @xmath21 and @xmath22 described above , enables one to do just that .
in particular , specifies the correct transformations for @xmath63 and @xmath64 , as transported by the quantum trajectories .
all spatial derivatives of arbitrary orders can then be obtained via spatial differentiation of although in the hard wall case , only the @xmath65 condition , @xmath66 is relevant , because all higher order derivatives are identically zero .
since the magnitudes of the @xmath67 and @xmath63 fields associated with a given quantum trajectory are unchanged as a result of the trajectory hop , implies that the incident and reflected flux values are the same ( apart from sign ) , and so the scattering system exhibits 100% reflection and zero transmission ( along each lm , the flux is invariant@xcite ) .
these basic facts of the hard wall system are of course well understood .
the point , though , is that we have now obtained the information in a time-_dependent _ quantum trajectory manner , rather than through the usual route of applying boundary conditions to time-_independent _ piecewise component functions . in other words , now refers to individual _
quantum trajectories _
, rather than to wavefunctions .
this shift of emphasis is very important , and leads to quite a number of conceptual and computational advantages .
for instance , the standard description of the hard wall stationary states would decompose these into plane wave components interpreted as `` incident '' and `` reflected '' waves .
this language suggests a process , or change over time
a state that is initially incident , at some later time is somehow transformed into a reflected state .
nothing in the standard description , however , would seem to render transparent the usage of such terminology , i.e. @xmath16 is stationary , and so the reflected and transmitted components are in fact _ both _
present for all times .
of course , a localized superposition of stationary states , i.e. a wavepacket , may well exhibit such an explicit transformation over the course of the time evolution , as such a state is decidedly non - stationary .
indeed , wavepackets are relied upon by the more rigorous formulations of scattering theory , in order to justify the use of terms such as `` reflected wave , '' even in a stationary context.@xcite such formulations , though certainly legitimate , seem always to require a clever use of limits , the subtle distinction between unitary and isometric transformations , and other esoteric mathematical tricks . on the other hand ,
the time - dependent bipolar quantum trajectory hopping picture presented above provides a physicality to such language that is immediately apparent . over the course of the time evolution , although the wavefunction as a whole is stationary , each individual _ trajectory _ is first incident from the left , then collides with the hard wall , and is subsequently reflected back towards the left ( i.e. towards @xmath27 ) .
the bipolar quantum trajectories are all classical , as the bipolar quantum potentials , @xmath32 , are zero everywhere except at the wall itself .
interference arises naturally from the superposition of the two lms i.e .
, from the trajectories that have already progressed to the point of reflecting , vs. those that have not reflected yet .
in contrast , since @xmath16 itself exhibits very substantial interference , and an infinite number of nodes , the traditional unipolar qtm treatment would be very ill - behaved , i.e. @xmath68 would oscillate wildly in the large @xmath69 limit , and @xmath70 would be numerically unstable near the nodes . apart from these important pragmatic drawbacks , the incident / reflected interpretation of the quantum trajectories would also be lost .
the bipolar quantum trajectory description of the hard wall system is very reminiscent of ray optics , as used to describe the reflection of electromagnetic waves off of a perfectly reflecting surface.@xcite indeed , much can be gained from applying a ray optics analogy to quantum scattering applications , especially where discontinuous potentials are concerned .
one can construct a simple gedankenexperiment as follows .
let @xmath71 denote some effective left edge of the system , well to the left of the interaction region . at some initial time
@xmath72 , all trajectories on the positive lm lying to the right of @xmath73 are _ ignored _ , as is the negative lm altogether .
one then evolves the retained trajectories over time , and monitors the contribution that just these trajectories make to the total wavefunction . in some respects ,
it is as if the point @xmath73 were serving as the initial wavefront for some incoming wave , that at @xmath72 had not yet reached the hard wall / reflecting surface . of course , if the actual wave were in fact truncated in this fashion , then the discontinuity in the field functions at the wavefront would result in a very non - trivial propagation over time , owing to the high - frequency components implicitly present . for actual waves , the precise nature of the wavefront
is known to have a tremendous impact on the resultant dynamics.@xcite we avoid such complicating details by always interpreting the `` actual wave '' to be the full stationary wave itself , i.e. the truncation is conceptual only . in the ray optics analogy ,
the above situation is like a source of light located at @xmath73 , which is suddenly `` turned on '' at @xmath72 .
it takes time for the wavefront to propagate to the reflecting surface , and additional time for the reflected wavefront to make its way back to @xmath74 .
prior to the latter point in time , the evolution of the truncated electromagnetic wave is decidedly _
non_-stationary ; afterwards however , a stationary wave is achieved , at least within the region of interest , @xmath75 , as the wavefront has by this stage propagated beyond this region .
the same qualitative comments apply to the bipolar quantum case , although of course the evolution equations are different . a similar prescription
may be used to achieve rudimentary `` wavepacket dynamics , '' even in the context of purely stationary states . instead of retaining _ all _ initial trajectories that lie to the left of @xmath73 , one retains only those that lie within some finite interval .
the resulting time evolution is analogous to a light source that is turned on at @xmath72 , and then turned off at some later time ( prior to when the wavefront arrives at the reflecting surface ) .
the initial `` wavepacket '' has uniform density , and moves with uniform speed towards the hard wall .
interference fringes then form after the foremost trajectories have been reflected onto the negative lm .
eventually , all trajectories within the interval are reflected , at which point interference ceases ( the nodes are `` healed''@xcite ) , uniform density is restored , and the reflected wave travels with uniform speed in the reverse direction , back towards the starting point @xmath73 .
qualitatively , this behavior is clearly similar to that undergone by actual wavepackets reflecting off of barrier potentials .
the ideas described above can be easily extended to more complicated discontinuous potential systems , such as up- and down - step potentials , and any combination of multiple steps , e.g. square barriers and square wells . in paper iii
, they will even be extended to arbitrary continuous potentials.@xcite in every case , the ray optics analogy from electromagnetic theory may also be extended accordingly .
this approach provides a useful perspective on global reflection and transmission in scattering systems , and in particular , demonstrates how such quantities may be obtained from a single , universal expression for local reflection and transmission .
we next consider the step potential system : v(x ) = [ steppot ] classically , this system exhibits 100% transmission if the trajectory energy is above the barrier ( i.e. @xmath76 ) , and 100% reflection if the trajectory energy is below the barrier ( @xmath77 ) .
quantum mechanically , all above barrier trajectories are found to exhibit partial reflection and partial transmission , although there is a general increase in transmission probability with increasing energy .
the below barrier quantum trajectories exhibit 100% reflection , as in the classical case ; however , they also manifest tunneling into the classically forbidden @xmath60 region .
thus even quantum mechanically , the the above and below barrier cases must be handled somewhat differently .
to begin with , we consider the above - barrier case .
note that the lm s are unbounded in either direction , i.e. the classically allowed region extends to both asymptotes , @xmath78 .
incoming trajectories can therefore originate from either asymptote , thus giving rise to two linearly independent solutions , @xmath16 .
this is in stark contrast to the hard wall system , for which incoming trajectories could only originate from @xmath79 , thus resulting in only one linearly independent solution for @xmath16 . in the standard time - independent picture ,
one starts with the four piecewise solutions , ( x ) = e^i p_a x/ ( x ) = e^i p_b x/ , [ steppieces ] where region @xmath80 corresponds to @xmath81,region @xmath82 to @xmath83 .
the momenta values are classical , i.e. p_a = p_b = . [ steppees ]
matching @xmath16 and @xmath84 boundary conditions at @xmath59 , and specifying asymptotic boundary conditions for @xmath16 , then enables a unique determination of the four complex coefficients @xmath85 and @xmath86 in ( x ) = . [ stepwhole ] in general ,
the solution coefficients depend on the particular stationary solution of interest . for the usual scattering convention of an incident wave incoming from the left ( fig .
[ stepabovefig ] ) the solutions are where @xmath87 and @xmath88 are respectively , reflection and transmission amplitudes . when flux is properly accounted for , the resultant reflection and transmission probababilities ( which add up to unity )
are given by p _
= |r|^2 ; p _ = |t|^2 .
[ steprefltrans ] note that above are correct for both an `` up - step '' and a `` down - step''i.e . for @xmath89 positive or negative .
we can also apply these equations to the `` opposite '' boundary conditions , i.e. to an incident wave incoming from the right , by simply transposing @xmath80 and @xmath82 subscripts , and @xmath90 and @xmath91 subscripts ( @xmath92 and @xmath93 are still positive ) .
this is important , because any stationary solution @xmath16 can be obtained as some linear superposition of left - incident and right - incident solutions . , as described in sec .
[ upstep ] .
solid and dashed lines represent real and imaginary contributions , respectively .
the dot - dashed line denotes the location of the step . ] regarding the time - dependent interpretation , it is evident that upon reaching the step discontinuity , left - incident trajectories must be partially reflected and partially transmitted .
the trajectory is suddenly split into two , one that continues to propagate along the positive lm for the transmitted @xmath82 region ( i.e. the @xmath94 lm ) and the other being instantaneously reflected down to the @xmath95 lm .
moreover , since probability carried by individual quantum trajectories is conserved,@xcite this splitting must be done in a manner that preserves both probability and flux . in other words ,
the local splitting of the trajectory at @xmath59 must correspond to , which is now regarded as a _ local _ condition , giving rise to local reflection and transmission amplitudes , @xmath87 and @xmath88 . for the present step potential case ,
these local quantities are directly related to the global @xmath96 and @xmath97 values via . for multiple step potentials ( sec . [ steps ] )
, the global expressions above [ ] no longer apply ; however , a local , time - dependent trajectory version of turn out to be correct .
such an expression , immediately applicable to all single and multiple step systems , can be written as follows : in the above equations , `` inc '' refers to any trajectory , locally incident on some particular step from some particular direction , which spawns both a locally reflected trajectory , `` refl , '' and a locally transmitted trajectory , `` trans '' .
the quantity @xmath98 is the ( positive ) momentum associated with the locally incident / reflected trajectory ; similarly , @xmath99 ( also positive ) is associated with the locally transmitted trajectory . for above - barrier
incident trajectories , note that the local reflection and transmission amplitudes are both real , thus ensuring the reality of @xmath63 and @xmath64 for the spawned trajectories .
returning to the step potential system , the ray optics picture can once again shed some interesting light .
the optical analog of the step is an interface between two media with different indices of refraction .
light incident on such an interface will partially reflect back towards the original source , and partially refract forwards into the new medium .
the refraction is completely analogous to the discontinuous change in momentum , @xmath100 , that suddenly occurs as one crosses the step ( fig .
[ trajfig ] ) . in any event , the ray optics gedankenexperiment described in sec .
[ hardwall ] can also be applied to the step potential system , in order to obtain a particular stationary solution @xmath16 with any desired boundary conditions .
for instance , suppose one is interesting in constructing the left - incident wave solution , i.e. that of . at @xmath72 , only the @xmath101 wave
is considered , and only those trajectories for which @xmath102 , as before . as the incident trajectories reach the step ,
two new waves are dynamically created from the spawned trajectories : a transmitted wave traveling to the right , and a reflected wave traveling to the left . a plot of the overall density @xmath103 so obtained will change over time , as the transmitted and reflected wavefronts propagate into their respective regions ( fig .
[ sbbelowfig ] ) .
eventually , however , these wavefronts will propagate beyond the region of interest , i.e. @xmath104 , where @xmath105 is the right edge of the region of interest .
when this occurs , the solution for @xmath16 obtained within the region of interest will be exactly equal to the stationary solution with the desired boundary condition .
as in the hard wall case , one can also perform step potential `` wavepacket dynamics '' by restricting consideration to just those initial @xmath101 trajectories lying within some coordinate interval .
the wavepacket will propagate towards the step with uniform density and speed .
as the first few trajectories hit the step , a uniform transmitted wave will be formed in the @xmath82 region . in the @xmath80 region , the sudden appearance of a @xmath106 wave will introduce interference wiggles in the overall density plot ( although no nodes _ per se _ , owing to partial reflection only ) . eventually ,
after all trajectories have progressed beyond the step , well - separated transmitted and reflected wavepackets emerge , propagating in their respective spaces and directions . there is no longer any interference in the @xmath80 region , as the incident wave is now gone , having been completely divided into the two final contributions . @xmath96 and @xmath97 values may be determined via monitors placed at @xmath73 and @xmath107 , either by integrating probability over time as the respective wavepackets travel through , or by recording the ( constant ) amplitude values @xmath87 and @xmath88 , and applying .
the case for which the incident trajectory energies are below @xmath89 requires special discussion . in this case , the classical lms and trajectories are confined to the @xmath80 region only ( i.e. to @xmath81 ) , as the entire @xmath82 region is classically forbidden . in the language of sec .
[ scattering ] , these below barrier states are therefore semi - bound , implying that there is only one linearly independent stationary solution , @xmath16 which without loss of generality , must be real - valued .
this in turn implies that the @xmath17 are complex conjugates of each other , as in the bound state case discussed in paper i. indeed , one option is to simply apply the paper i decomposition to such problems .
this approach is discussed in detail in the appendix , wherein it is shown to provide a natural extension of classical trajectories into the tunneling region .
, as described in sec . [ upstep ] .
solid and dashed lines represent real and imaginary contributions , respectively , for the analytical solution .
squares and circles denote corresponding numerical results .
the shaded box represents the tunneling region . ] on the other hand , the trajectory hopping - based decomposition scheme offers a different , but also very natural means to accomplish the same task which has the added advantage that all bipolar quantum potentials _ vanish _ , except at @xmath59 .
the idea is simply to treat all expressions in sec .
[ stepabove ] as being literally correct for the below barrier case as well , with the understanding that the requisite quantities need no longer be real .
in particular , @xmath108 [ ] becomes pure positive imaginary , implying that the transmitted trajectories `` turn a corner '' in the complex plane , and start heading off in the positive imaginary direction , with speed @xmath109 ( fig .
[ complexfig ] ) . along this path
, the transmitted wave is an ordinary plane wave ; however , when analytically continued to the real axis in the @xmath60 region ( via a @xmath110 clockwise rotation in the complex plane ) , the familiar exponentially damped form results ( fig .
[ stepbelowfig ] ) . for the reflected wave , states that the reflected `` phase '' remains unchanged .
however , @xmath111 is now _ complex _ , leading to an effective phase shift of @xmath112 , where @xmath113 is defined in the appendix [ ] . for localized wavepackets , a physical significance can be attributed to this phase shift , in both quantum mechanics and electromagnetic theory ; it is the source of the goos - hnchen effect,@xcite a time delay observed in conjunction with total internal reflection .
consequently , in the time - dependent wavepacket context , it may be more appropriate to associate the phase shift with _ time _
, rather than with @xmath64 or @xmath63specifically , with the delay time needed to accrue sufficient action so as to compensate for the shift .
for stationary states , however , such a time delay would be inconsequential , because all trajectories are identical apart from overall phase .
consequently , we do not consider such time delays explicitly in this paper , though we will return to this issue in future publications .
the most interesting case is that for which there are multiple discontinuities , occurring at arbitrary locations @xmath114 ( with @xmath115 ) , and dividing up configuration space into @xmath116 regions , labeled @xmath80 , @xmath82 , @xmath117 , etc . in each region
, the potential energy has a different constant value , i.e. @xmath118 in region @xmath80 , etc . from an optics point of view , this system is analogous to a stack of different materials , each with its own thickness , and index of refraction
. our primary focus in this paper will be square barrier / well systems for which @xmath119 , and @xmath120 .
however , all of the present analysis extends to the more general case described above . in the standard time - independent picture ,
the solution is obtained via a straightforward generalization of eqs .
( [ steppieces ] ) , ( [ steppees ] ) , and ( [ stepwhole ] ) . however , even when comparable left - incident boundary conditions are specified as in sec .
[ stepabove]i.e .
@xmath121 , and ( for @xmath119 ) @xmath122the remaining coefficient values are fundamentally different from those of the single - step case . to begin with
, only the @xmath123th step exhibits the characteristics of a ( locally ) left - incident single - step solution ; all other steps involve four non - zero coefficients , corresponding locally to some superposition of left- and right - incident waves . even more importantly , however , the expressions for the coefficient values as a function of system parameters _ in no way _ resembles ; in particular , these now depend explicitly on the @xmath114 values , as well as on @xmath124 , @xmath125 , etc .
the same is also true for the global @xmath96 and @xmath97 expressions , as compared with .
it is this dependence on the other steps that gives rise to the global nature of the time - independent solutions ; i.e. the coefficient values at one step depend in principle on the properties of all of the other steps , no matter how far away these might be located .
consequently , a reflection probability as obtained from the @xmath126 value associated with the first , @xmath127 step , can not be determined without extending the analysis out to the final step at @xmath128 , in the standard time - independent picture . on the other hand , a primary goal of the time-_dependent _
approach is to construct a completely _ local _ theory , for which local reflection and transmission amplitudes associated with any given trajectory , as it encounters a given step @xmath69 , depend _ only _ on the properties of the @xmath69th step ( i.e. on @xmath114 , and on the @xmath67 or @xmath129 values to the immediate left and right of @xmath114 ) .
in fact , from the point of view of the given trajectory , it must be immaterial whether the potential contains other steps or not implying that the correct local relations for the spawned trajectories , if they exist at all , must be exactly those already specified in .
how is it possible that for stationary wavefunctions , whose time evolution is presumably trivial , an inherently global problem can be converted to a local one , simply by switching from a time - independent to a time - dependent perspective ? this is because of the bipolar decomposition , which provides each step with not one , but two sets of incident trajectories , one from the left , and one from the right . when there are multiple steps , not only does this result in a non - trivial superposition for the resultant locally reflected and transmitted waves , but the trajectories themselves are subject to multiple spawnings , which effectively enable them to traverse back and forth over the same regions of configuration space an arbitrary number of times ( fig .
[ trajfig ] ) .
this crucial feature ultimately gives rise to the rich global scattering behavior observed even in two - step systems .
however , it is wholly missed by any time - independent treatment , even a bipolar one , which can only summarize the net superposition of all left - traveling and right - traveling waves .
, as described in sec . [ squarebarrier ] .
one trajectory in five is indicated in the figure .
the black / gray solid lines indicate positive / negative lm trajectories , respectively .
the open circles represent recombination points .
the dashed lines denote the two barrier edges , @xmath130 and @xmath131 . ]
we now discuss how the local time - dependent theory described above gives rise to the correct stationary solutions , which is readily understood by invoking the ray optics description introduced earlier . for simplicity and definiteness
, we consider only the square potential case , which is optically analogous to say , a single pane of glass surrounded by vacuum . if a single step gives rise to a single reflection , then two steps , like a pair of mirrors , results in an _
infinite _ number of reflections .
the same is true of a pane of glass , within which a single beam of light will be reflected back and forth at the edges an arbitrary number of times . of course
, these reflections are not perfect ; a portion of the incident flux always escapes as transmission into the surrounding vacuum .
consequently , each successive internal reflection is exponentially damped , in accord with .
if the globally incident wave is incoming from the left , then at @xmath132 , there are two contributions to @xmath133 .
one contribution is the portion of the left - incident @xmath134 wave that is _ locally transmitted _ through the first step .
apart from a phase factor , the resultant @xmath135 value would be given by if this were the only contribution .
however , there is also a contribution that arises from the _ locally reflected _ part of the right - incident wave , @xmath136 .
this contribution is zero for a single step system , but of course non - zero in the multiple step case .
although the second contributing wave is right - incident , we can still use to compute the contribution to @xmath133 , as discussed in sec .
[ stepabove ] . for the second , @xmath137 step at @xmath138
, there are only left - incident waves ; consequently , @xmath139 and @xmath136 are obtained from a single source each , i.e. @xmath133 ( fig . [ trajfig ] ) .
, for the @xmath76 square barrier problem , as computed using the numerical algorithm of sec . [ numerics ] .
the shaded box represents the barrier region .
all units are atomic .
the evolving discontinuities found at intermediate times in these curves denote wavefronts for the newly created reflected / transmitted components . over time , the magnitudes of these discontinuities ( i.e. corrections ) become arbitrarily small , signifying that numerical convergence to the left - incident stationary solution has been achieved . ]
the above description refers to the stationary state result , obtained by our gedankenexperiment in the large time limit only . in practice
this result would be achieved in stages . as in the previous examples
, we imagine that at time @xmath72 , one retains only those trajectories for which @xmath140 .
this one - sided trajectory restriction is somewhat analogous to continuous wave cavity ring - down spectroscopy.@xcite when the wavefront first hits the first interface at @xmath141 , there is partial reflection and transmission , exactly identical to what would happen for a single step system .
the reflected wavefront propagates beyond the left edge of the region of interest at @xmath74 , and for some time , the reflected amplitude passing through this left edge is constant .
the initially transmitted wavefront eventually reaches the second step at @xmath142 ( i.e. the far side of the pane of glass ) , leading to a second transmission into the @xmath117 region , and a second reflection back through the @xmath82 region .
eventually , the second transmitted wavefront reaches the right edge of interest at @xmath143 , after which the transmitted amplitude remains constant for some time .
neither the globally transmitted nor reflected amplitudes for the times indicated above , as determined via monitors at @xmath74 and @xmath144 respectively , are correct .
however , we have not yet described the steady state solution .
to do so requires an accounting of the second reflected wavefront , which eventually reaches the @xmath141 step again , this time incident from the right .
the resultant locally transmitted wave becomes an instantaneous second contribution to @xmath145 , and the locally reflected wave plays the same role for @xmath133 .
these new contributions give rise to discontinuities in these waves , that subsequently propagate to the left and right , respectively ( fig.[sbabovefig ] ) .
the new @xmath145 wave discontinuity eventually reaches @xmath74 , where it is recorded by the monitor , giving rise to a sudden change in the reflection probability value .
the @xmath133 discontinuity propagates to the second step , where it spawns new discontinuities in @xmath139 and @xmath136 .
the former constitutes the border between first- and second - order transmitted waves , registered at sufficiently later time by the monitor at @xmath146 .
the latter , second - order @xmath136 wave heads back towards the first step , to give rise to third - order waves , with commensurate discontinuities , etc . in principle
, this process continues indefinitely , resulting over time in global transmitted and reflected waves of arbitrarily high order .
however , and the relation @xmath147 ensure that the result converges to a stationary solution exponentially quickly .
moreover , since @xmath148 is necessarily zero throughout this process , it is clear that the stationary state that is converged to is indeed the one corresponding to the desired boundary condition of a globally incident wave that is incoming from the left .
note that in an actual optical system as described above , the spatial dimensionality is three rather than one , and the incident wave would usually be taken at some angle to the normal .
if in addition , the beam has a finite width , then one would observe separate reflected beams for each order , of exponentially decreasing brightness .
the one - dimensional quantum case , however , is analogous to a normal incident beam , for which all orders of reflection are superposed .
in addition to providing a pedagogical understanding of the dynamics that is very much analogous to the optical example provided , the picture above also suggests a practical numerical method that may be used to obtain stationary scattering states of any desired boundary condition ( via superposition of globally left- and right - incident wave solutions , obtained independently ) .
note that the `` wavepacket dynamics '' version of the ray optics analogy may also be applied . in this case , the resultant initial square wavepacket is somewhat reminiscent of pulsed wave cavity ring - down spectroscopy.@xcite once the wavepacket has penetrated the middle region @xmath82
( i.e. the pane of glass ) , it reflects back and forth between the two edges , with each reflection giving rise to a left- or right - propagating outgoing square wavepacket in region @xmath80 or @xmath117 , and a temporary interference pattern in region @xmath82 .
the amplitude of the central wavepacket dissipates exponentially in time .
all of this complicated behavior is indeed qualitatively observed in actual wavepacket dynamics for such systems , but in the present context , is reconstructed entirely from a single stationary state .
in this section , we discuss several remaining issues pertaining to the numerical methods used to generate and propagate the various bipolar component waves , for the examples discussed in secs .
[ complicated ] and [ results ] .
for all of these examples , the numerical algorithm used corresponds to the gedankenexperiment with one - sided truncation , i.e. to continuous wave cavity ring - down .
in essence , this consists of just two basic operations : ( 1 ) each piece - wise bipolar component of the wavefunction [ i.e. @xmath149 , @xmath150 , etc . ]
is independently propagated in time over its appropriate region of space , using the standard qhems and qtms ; ( 2 ) whenever a trajectory reaches a turning point , it is immediately deleted , and replaced with two new trajectories , spawned in the appropriate locally transmitted and reflected component lms . the first operation above , i.e. qtm propagation of the wavefunction components , is very straightforward .
note that for simplicity , we have throughout this paper used time - independent expressions for @xmath16 and its components , but in reality these evolve over time even for stationary states , via @xmath151 .
we have therefore been rather lax in distinguishing hamilton s principle function from hamilton s characteristic function , although from a trajectory standpoint , it is always the former that is implied .
since each component is stationary in its own right , the time evolution of the hydrodynamic fields is governed by the quantum stationary hamilton - jacobi equation ( qshje ) , rather than the qhje .
moreover , the fact that the piecewise @xmath63 is constant implies that the component quantum potentials are _ zero _ , resulting in classical hje s and trajectories .
these conclusions are trivially correct for the present paper , for which all components are plane waves ; however , the arguments also extend to arbitrary continuous potential systems.@xcite from a numerical perspective , the use of classical trajectories offers many advantages over a conventional qtm propagation . to begin with ,
the trajectories themselves are always smooth if @xmath13 is smooth , resulting in far fewer trajectories and larger time steps than would otherwise be the case .
even more importantly , however , since the quantum potential is not required , there is no need to compute on - the - fly numerical spatial derivatives of the local hydrodynamic fields .
consequently , for a given component , the trajectories are completely independent and need not communicate again resulting in fewer of them .
indeed , it is possible to perform an essentially exact computation using only a _ single trajectory per wavefunction component_. this feature is particularly important for the very frontmost trajectory of the initial ensemble , which for brief periods at later times , will ( via spawning ) come to be the _ only _ trajectory to occupy a given component lm .
the subsequent evolution of these lone trajectories does not require the presence of nearby trajectories .
the spawning of new trajectories , i.e. operation ( 2 ) above , also bears further discussion . in principle , this is always achieved via application of . for the above barrier case ,
@xmath87 and @xmath88 are both real , ensuring the reality of @xmath63 and @xmath64 for the spawned trajectories
although in the case of a down step , @xmath152 , resulting in a negative @xmath111 .
this is in accord with the conventions discussed in paper i. however , in this paper , we find it numerically convenient to adopt the more usual @xmath153 convention .
thus , if yields a negative @xmath111 , it is replaced with @xmath154 , and @xmath155 is added to @xmath156 . a similar , but more complicated modification is also applied to the below - barrier trajectories , for which yield complex amplitudes . in this case , the phase shift is @xmath157 , as discussed in sec .
[ stepbelow ] and the appendix . for a single step system , the algorithm is now essentially complete . at the initial time
@xmath72 , a variable number of particles ( or synonymously , grid points ) are distributed uniformly along the @xmath134 manifold , to the left of @xmath158 .
the extent of these points must be large enough that at the end of the propagation , there are still @xmath134 grid points that have not yet reached @xmath73 .
the grid spacing is mostly arbitrary , but must be small enough that at sufficiently later times , there is always at least one trajectory per component lm . the propagation is considered complete when the reflected and transmitted wavefronts travel beyond @xmath73 and @xmath107 , respectively . for multiple step systems ,
the situation is similar , but somewhat more complex .
the primary new feature is the _ recombination _ of wavefunction components arising from two sources , i.e. from two locally incident waves coming from opposite directions .
the present algorithm would seem to yield _ two _ subcomponent wavefunctions for every component , each with its own set of trajectories .
if left `` unchecked , '' this would lead to undesirable further multifurcations for higher orders / later times .
a simple solution would be to propagate each subcomponent long enough that there is at least one trajectory for each , then extrapolate the corresponding subcomponent wavefunctions to a common position , where a new trajectory is constructed for the superposed component wavefunction , which is then propagated in lieu of the subcomponent trajectories .
this requires dynamical fitting ( see below ) , or at the very least , extrapolation .
although these numerical operations would be very stable in the present context , to rule these out altogether as sources of error in sec .
[ results ] , we have adopted a much simpler approach i.e . the grid spacing is chosen such that trajectories from the two component waves incident on a given step always arrive at the same time ( fig .
[ trajfig ] ) .
the corresponding subcomponent wavefunction values are then simply added together when forming the spawned trajectory .
adopting once again the @xmath153 convention for the superposed component wave , @xmath159 , the corresponding field values are then obtained via @xmath160 and @xmath161 . as discussed in sec .
[ steps ] , multiple step systems allow for infinite reflections that perpetually modify @xmath103 , in principle for all time . in practice however , there is exponential convergence within the region of interest , @xmath162 , so that one would not run the calculation indefinitely , but only until the desired accuracy is reached .
accurate `` error bars '' on the computed global @xmath96 and @xmath97 values are conveniently provided by the magnitudes of the most recent discontinuous jumps as recorded by the monitors at @xmath73 and @xmath107 .
note that the number of digits of accuracy scales only _ linearly _ with propagation time .
however , the rate of convergence depends on the energy value . near the barrier height ,
in particular , convergence may take quite a long time , as the exponent is close to zero .
for all other energies , only a few `` cycles '' should be required , depending on the level of accuracy desired .
if in addition to reflection and transmission probabilities , the actual stationary solution over the region of interest is also desired , then it is necessary to reconstruct @xmath16 .
this is obtained from the final grid , after the propagation is finished , using a multiple step generalization of .
the first step is to reconstruct the component wavefunctions @xmath149 , @xmath150 , etc .
, via interpolation or fitting of the hydrodynamic field values from the corresponding dynamical grid points onto a much finer common grid ( used e.g. for plotting purposes ) .
the second step is to linearly superpose the @xmath163 components onto the plotting grid , and to assemble the pieces together over the coordinate range of interest . for the discontinuous systems considered here
, the number of dynamical grid points per component can be as small as one i.e .
much smaller , even , than the number of wavelengths ! to our knowledge , such performance has never been achieved previously by a qtm ; however , it does require that the plotting grid be much finer than the dynamical grid , e.g. at least several points per wavelength , in order to adequately represent the interference fringes of the the superposed solution , @xmath16 .
in this section , we apply the numerical algorithm previously described to three different applications : the up - step potential , the square barrier , and the square well .
the first system considered is the up - step potential , i.e. with @xmath164 .
since there are no multiple reflections , this is in principle a trivial application for the current algorithm ; it therefore serves as a useful numerical test . both above barrier ( sec .
[ stepabove ] ) and below barrier ( sec .
[ stepbelow ] ) energies are considered .
we choose molecular - like values for the constants , i.e. @xmath165 hartree , and @xmath166 a.u .
the left and right edges of the region of interest are taken to be @xmath167 a.u . and
@xmath168 a.u . ,
respectively . at the initial time , @xmath72 , 51 trajectory grid points are distributed uniformly over the interval @xmath169 ( grid spacing of @xmath170 a.u . ) .
this number is far greater than what would be needed for dynamical purposes , but is chosen so as to avoid construction of a separate plotting grid ( sec .
[ numerics ] ) .
the hydrodynamic field functions for the initial @xmath171 wavepacket over the above interval are taken to be @xmath172 a.u.@xmath173 and @xmath174 . for the above barrier calculation
, the energy @xmath175 hartree was used . the trajectory propagation and termination
were performed exactly as described in sec .
[ numerics ] .
the real and imaginary parts of all three resultant wavefunction components [ i.e. @xmath149 and @xmath176 $ ] at the final time , @xmath177 a.u . are presented in fig .
[ stepabovefig ] .
all three components exhibit the desired plane wave behavior , e.g. no interference is evident within a given component . the resultant @xmath16 does exhibit interference in the @xmath80 region , however , arising from the superposition of @xmath101 and @xmath106 . for the below barrier calculation ,
the system was given an energy equal to one half of the barrier height , i.e. @xmath178 . as per the discussion in secs .
[ stepbelow ] and [ numerics ] , tunneling into the forbidden region @xmath82 is achieved , not through a quantum potential , but via analytic continuation . at sufficiently large time
( @xmath179 a.u . ) the final wavefunction is reconstructed from the components , i.e. @xmath180 , and @xmath181 .
the real and imaginary parts of the reconstructed wavefunction are presented in fig .
[ stepbelowfig ] . in the figure ,
squares and circles denote the numerical results obtained via the present algorithm , whereas the solid and dashed lines represent the well - known analytic solutions .
the agreement is essentially exact .
note that the real and imaginary parts are in phase throughout the coordinate range
, @xmath182 is a constant , so apart from a phase factor , @xmath16 is real .
note also that the tunneling region exhibits the desired exponential decay .
the second system considered is the square barrier .
this is a two - step potential ( @xmath119 ) , with @xmath183 , and @xmath184 .
the two steps comprise the left and right edges of the barrier , at @xmath185 and @xmath186 , respectively .
the constants are chosen as follows : @xmath187 hartree ; @xmath166 a.u . ;
@xmath188 ; @xmath189 a.u . , @xmath190 a.u .
initially , 75 trajectory grid points are distributed uniformly over the interval @xmath191 ( grid spacing of @xmath192 a.u . ) , which again , is far more than are dynamically required . the same initial hydrodynamic field functions are used as in sec .
[ upstep ] . both above barrier ( @xmath76 ) and below barrier ( @xmath77 ) energies are considered . for the above barrier case , @xmath193 hartree .
once again , the trajectory propagation and termination were performed exactly as described in sec .
[ numerics ] . in order to converge @xmath97 to @xmath194 , a propagation time of @xmath195 a.u . was required .
this corresponds to 3 complete cycles , i.e. a 3rd - order calculation .
figure [ trajfig ] is a plot of the quantum trajectories for this calculation , in which every fifth trajectory for each of the five component wavefunctions is indicated .
trajectory spawning at the two steps is very clearly evident , as is recombination of pairs of incident waves ( indicated by circles ) . on the whole , this figure demonstrates all of the anticipated analogues with ray optics , i.e. parallel trajectories , reflection and refraction . in fig .
[ sbabovefig ] , the time evolution of the superposition state @xmath196 is represented , via snapshots of the real and imaginary parts at seven different times . at @xmath72 a.u . , the incident wavefront is located at @xmath197 a.u . by @xmath198 a.u . , the wavefront has spawned @xmath199 and @xmath200 trajectories ; the former gives rise to the kink ( really a discontinuity ) somewhat to the left of the first step .
by @xmath177 a.u . and @xmath201 a.u .
, the @xmath199 wavefront has moved outside the region of interest , though the @xmath200 wavefront has not quite reached the second step .
after it does so , two new wavefronts are propagated along the @xmath202 and @xmath203 lms ( e.g. @xmath204 a.u . ) , the former of which propagates beyond the region of interest by @xmath205 a.u .
subsequent discontinuity magnitudes become exponentially smaller , so that at sufficiently large time ( i.e. @xmath206 a.u . ) , the resultant @xmath16 has converged to the correct stationary solution . for the below barrier case , @xmath207 hartree , @xmath208 a.u . , and the other parameters are as above except @xmath209 a.u . within the barrier ,
there are in principle an arbitrary number of reflections back and forth as before .
however , substantial amplitude loss occurs due to tunneling , in addition to partial reflection , as a result of which fewer cycles are required in order to achieve the same @xmath194 level of convergence ( @xmath210 a.u . , or 2 cycles ) .
[ complexfig ] indicates how the tunneling dynamics is achieved .
after the wavefront hits the first step , the transmitted @xmath211 wave is propagated along the imaginary axis ( @xmath212 ) , until the point @xmath213 is reached . when this occurs , it is necessary to analytically continue @xmath211 down to the real axis , in order to compute amplitudes for the new @xmath202 and @xmath203 trajectories that are spawned at the second step .
the latter component propagates in the ( negative ) imaginary direction , along @xmath214 , until @xmath215 , at which point analytic continuation is once more applied ( this time for the first step ) , and the pattern repeated .
square barrier problem , as discussed in secs .
[ numerics ] and [ squarebarrier ] . ]
seven snapshots of the superposition density , @xmath103 , are displayed in fig .
[ sbbelowfig ] . the initial density
equal to just the @xmath216 density is uniform .
after the wavefront encounters the first step , a reflected @xmath199 emerges , giving rise to clearly evident interference in the region @xmath80 .
the @xmath217 wave is at this stage perfectly exponentially damped . upon encountering the second step , a second contribution
, @xmath203 emerges ; however , this first - order correction is already extremely small , owing to the large amount of tunneling that has occurred . the global transmitted wave , @xmath202 , though small , is clearly seen to have uniform density .
in addition to the two detailed trajectory calculations described above , we computed @xmath96 and @xmath97 for a large range of @xmath218 and @xmath18 values , so as to fully explore ( without loss of generality ) the entire range of the square barrier problem . the numerical results are presented , and compared with known analytical values,@xcite in fig . [ sbtransreflfig ] .
two aspects of this study bear comment .
first , for all @xmath218 and @xmath18 values considered , the computed @xmath96 and @xmath97 values agree with the exact values to within an error comparable to that predicted by the level of numerical convergence .
in particular , the oscillatory energy dependence is perfectly reproduced .
second , the closer the barrier peak is approached from either above or below in energy , the longer the time required to achieve a given level of convergence , as predicted in sec .
[ numerics ] .
in particular , for the calculations closest to the barrier peak , 5 - 7 cycles were required in order to approximately maintain a @xmath194 convergence of the transmission probability .
although a greater number of particles are required in this case , this poses no great limitation in practice , since one would presumably never require a calculation precisely at the peak energy .
, for the @xmath77 square barrier problem , as computed using the numerical algorithm of sec . [ numerics ] .
the shaded box represents the barrier region .
all units are atomic . note
that interference manifests in the incident ( left ) region only after some incident trajectories have struck the left barrier edge , causing reflected trajectories to be created ( i.e. just prior to @xmath219 , initially ) .
the most advanced reflected trajectory defines the reflected wavefront , manifesting as the left - moving discontinuity , e.g. at @xmath219 and @xmath220 . ] , as discussed in sec .
[ squarebarrier ] .
solid lines denote analytical results ; open / closed circles denote numerical results , as obtained via algorithm of sec .
[ numerics ] .
the vertical dot - dashed lines represent the barrier height , i.e. @xmath221 . ] as the final system , we consider the square - well potential , i.e. the square barrier but with @xmath222 . in the scattering state context
, there is no tunneling for this system , but in other respects it resembles the square barrier . from an optics point of view
, the square well corresponds to a central medium with larger index of refraction than its surroundings , whereas the square barrier corresponds to a smaller index of refraction , giving rise to the possibility of total internal reflection ( i.e. tunneling ) .
the parameters are as in sec .
[ squarebarrier ] , except that @xmath223 hartree , and three different @xmath218 values are considered : @xmath224 a.u .
, @xmath225 a.u . , and @xmath226 a.u .
as the time evolution and trajectory pictures are similar to those of the previous sections , we focus only on the @xmath96 and @xmath97 calculations , for which once again , a large range of energies was considered ( @xmath227 hartree ) .
the number of initial trajectories ranged from 50 to 200 , for the highest to the lowest energies , respectively . the computed transmission / reflection probabilities
were again converged to @xmath194 . the energy - resolved reflection and transmission probabilities
are presented in fig .
[ swtransreflfig ] . , as discussed in sec .
[ squarewell ] .
solid lines denote analytical results ; open / closed circles denote numerical results , as obtained via algorithm of sec . [ numerics ] . ] as in the square barrier case , excellent agreement is achieved with the exact analytical results , i.e. on the order of the level of convergence .
this is true despite the fact that the square well energy curves are decidedly more oscillatory than the square barrier curves particularly for wide barriers , for which the @xmath18 dependence is very sensitive indeed .
one particularly important feature exhibited by the exact curves is the so - called ramsauer - townsend effect,@xcite i.e. the phenomenon of 100% transmission and zero reflection , even at very low energies .
this occurs when @xmath228 , and may be regarded as a purely quantum mechanical resonance phenomenon . yet it is reproduced perfectly here in the bipolar decomposition , using _
classical trajectories_. indeed , the bipolar picture provides an interesting physical explanation , i.e. the right - incident and left - incident waves of the first step give rise to spawned contributions to @xmath199 that exactly cancel each other out via destructive interference .
as described in paper i , the equation is linear , yet the equivalent qhem obtained via substitution of the madelung - bohm ansatz into the equation are not .
this aspect of bohmian mechanics suggests that it can be beneficial , both from a pedagogical and a computational perspective , to apply a suitable bifurcation ( or `` multifurcation '' ) to the wavefunction prior to applying the qhem .
indeed , following the paper i cpwm bipolar decomposition for 1d stationary bound states,@xcite the quantum trajectories become more well - behaved and classical - like , in precisely the limit in which there are more nodes , and the usual unipolar calculation breaks down .
moreover , the resultant component lms admit a natural physical interpretation in terms of the corresponding semiclassical lms . in the generalization to the stationary scattering states considered here , a somewhat different bipolar decomposition is found to be required .
the new decomposition is still unique , at least for discontinuous potentials .
however , the resultant components @xmath17 are no longer solutions to the equation in their own right , as a result of which their time evolution is coupled . the new scheme
though fundamentally different from the old one nevertheless bears a correspondence to a modified version of semiclassical theory appropriate for scattering systems .
curiously , this semiclassical modification is _ not _ simply a higher order treatment in @xmath229;@xcite if it were , the corresponding exact quantum modification considered here would not exist . in any event , the new decomposition also gives rise to its own physical interpretation , specific to the scattering context .
in particular , for right - incident boundary conditions , the left and right asymptotes of @xmath21 respectively represent incident and transmitted waves , whereas the left asymptote of @xmath22 represents the reflected wave [ the right asymptote of @xmath22 approaches zero ] .
that these conceptually useful asymptotic bipolar assignments found even in the most elementary treatments of scattering may be extended _ throughout configuration space _ [ even for continuous potentials ( paper iii ) ] represents an important leap forward , especially for qtms .
another pedagogically and numerically useful development from this approach is the inherently time - dependent ray optics interpretation that naturally arises , particularly in the discontinuous potential context .
the ray optics approach is anticipated to be a relevant guiding force in subsequent generalizations of the present methodology , i.e. to continuous , multidimensional potential systems , and as per the discussion in secs .
[ hardwall ] and [ stepabove]for non - stationary wavepacket dynamics .
this approach also provides a much simpler trajectory - based explanation of scattering terminology as applied in a stationary context
why the `` reflected wave '' is so - called , despite being present from the earliest times than those traditionally used.@xcite the discontinuous potential applications considered in this paper hard wall ( sec .
[ hardwall ] ) , step potential ( secs .
[ stepabove ] , [ stepbelow ] , and [ upstep ] ) , and square barrier / well ( secs .
[ steps ] , [ squarebarrier ] , and [ squarewell])are significant for several reasons . to begin with ,
these are the first discontinuous applications of a genuine qtm calculation that have ever been performed , to the authors knowledge .
the singular derivatives associated with discontinuous potential functions would wreak havoc with standard numerical differentiation routines .
second , the use of a time-_dependent _ method for stationary , or time-_independent _ , applications , is also significant .
ordinarily , the time dependence of stationary states is regarded as trivial . in the present context , this is true in a sense for the hard wall and step potential systems , because the correct answer is `` built in '' the method itself . for multiple step systems , however , the dynamical truncated wave approach , i.e. the gedankenexperiment introduced in sec .
[ hardwall ] and further developed in later sections , yields decidedly nontrivial results .
in particular , the algorithm uses only _ single step _ scattering coefficients to obtain global scattering quantities for _ multiple step _ systems . in effect , the time dependent nature of this approach allows computation of _ global _ properties using a completely _ local _ method . not only were exact quantum results obtained for a full range of system parameters , but the numerical resources necessary to achieve this
the number of trajectories and time steps were decidedly minimal .
indeed , the algorithm lives up to the promise made in paper i , of performing an accurate quantum calculation with _ fewer trajectories than nodes_a prospect virtually unheard of in a unipolar context . in future publications , we will naturally attempt to generalize the methodology described here and in paper iii , for the type of multidimensional time - dependent wavepacket dynamics relevant to chemical physics applications . in this context
, the scattering version of the cpwm decomposition developed here is an absolutely essential first step , as reactive scattering is the underpinning of all chemical reactions .
additional discussion and motivation will be provided in paper iii .
this work was supported by awards from the welch foundation ( d-1523 ) and research corporation .
the authors would like to acknowledge robert e. wyatt and eric r. bittner for many stimulating discussions .
david j. tannor and john c. tully are also acknowledged .
as discussed in secs . [ scattering ] , [ hardwall ] , and [ stepbelow ] , semi - bound stationary states in 1d are bounded on one side only , as a result of which they are real - valued and singly - degenerate , like bound states .
consequently , they are amenable to the cpwm bipolar decomposition scheme introduced in paper i. in this appendix , we apply this decomposition to two semi - bound systems : the hard wall system , and the below - barrier up - step system . from ref .
, the most general bipolar decomposition of a hard wall stationary state corresponding to sec . [
bipolar ] condition ( 1 ) only is found to satisfy - = , where @xmath230 , and @xmath231 is obtained from @xmath232 via ( without `` sc '' subscripts ) .
the arbitrary parameters @xmath51 and @xmath82 are the invariant flux and median action parameters associated with conditions ( 2 ) and ( 3 ) , respectively,@xcite although the definition of @xmath51 has been changed slightly to account for the scattering normalization convention , @xmath233 .
note that _ only _ the semiclassical values for these parameters yields a solution that satisfies the correspondence principle in the large action ( i.e. @xmath69 ) limit .
in particular , the choice @xmath234 and @xmath235 yields the desired semiclassical result , @xmath236 ; all other choices exhibit undesirable oscillatory behavior in @xmath33 , @xmath45 , and @xmath32 . for the hard wall system considered above which is just the special case of the up - step potential in the limit @xmath237exact agreement is achieved between semiclassical and quantum lm s in the @xmath58 region .
this is the only region of interest for the hard wall system ; however , for finite @xmath89 values i.e . for general below - barrier up - step stationary states
there is of course also tunneling into the forbidden region , which must be accounted for .
the paper i bipolar decomposition therefore results in lm s that span the _ entire _ coordinate range @xmath238 .
these lms are given by the following analytical expression : p(x ) = , [ ptun ] where @xmath113 is given by = k = , [ deltaeqn ] and = .
[ kappaeqn ] the @xmath239 lm function of is continuous everywhere , including at the potential discontinuity at @xmath59 . in the @xmath80 region
, it agrees exactly with the semiclassical solution ; in the @xmath82 region , it decays exponentially to zero .
the paper i approach thus yields a very natural way to extend trajectories into the tunneling region .
note that the quantum potential in this region is not zero ; indeed , it exhibits a discontinuity at @xmath59 that exactly balances that of @xmath13 itself , so that the bipolar modified potential is continuous across the step . unlike the above - barrier case , the paper i solution does _ not _ manifest oscillatory behavior in the large action limit , and so this approach would at first glance appear to be ideal .
there are two reasons , however , why it is not pursued here .
the first reason is that @xmath240 diverges asymptotically as @xmath28 , which according to preliminary numerical investigations , appears to lead to numerical instabilities for completely qtm - based propagation schemes .
second , if the barrier were to fall off again at larger @xmath241 values , so that the tunneling region were finite , then the asymptotic behavior would be once again undesirably oscillatory
. this would be the case , for example , for the below - barrier energies of the square barrier system of sec .
[ squarebarrier ] .
|
in a previous paper [ j. chem .
phys . * 121 * 4501 ( 2004 ) ] a unique bipolar decomposition , @xmath0 was presented for stationary bound states @xmath1 of the one - dimensional equation , such that the components @xmath2 and @xmath3 approach their semiclassical wkb analogs in the large action limit . moreover , by applying the madelung - bohm ansatz to the components rather than to @xmath1 itself , the resultant bipolar bohmian mechanical formulation satisfies the correspondence principle . as a result ,
the bipolar quantum trajectories are classical - like and well - behaved , even when @xmath1 has many nodes , or is wildly oscillatory . in this paper ,
the previous decomposition scheme is modified in order to achieve the same desirable properties for stationary scattering states .
discontinuous potential systems are considered ( hard wall , step , square barrier / well ) , for which the bipolar quantum potential is found to be _
zero _ everywhere , except at the discontinuities .
this approach leads to an exact numerical method for computing stationary scattering states of any desired boundary conditions , and reflection and transmission probabilities .
the continuous potential case will be considered in a future publication .
|
the stellar initial mass function ( imf ) , @xmath1 , where @xmath2 is the stellar mass , is the parent distribution function of the masses of stars formed in _ one _ event . here
, the number of stars in the mass interval @xmath3 is @xmath4 .
salpeter ( 1955 ) inferred the imf from solar - neighbourhood star - counts applying corrections for stellar evolution and galactic - disk structure finding @xmath5 , for @xmath6 .
miller & scalo ( 1979 ) and scalo ( 1986 ) derived the imf for @xmath7 using better data and a more sophisticated analysis establishing that the imf flattens or turns - over at small masses .
modern studies of solar - neighbourhood star - count data which also apply detailed corrections for unknown multiple systems in the star - counts , confirm that @xmath8 for @xmath9 ( kroupa , tout & gilmore 1993 ; kroupa 1995b ; reid , gizis & hawley 2002 ) .
using spectroscopic star - by - star observations , massey ( 2003 ) reports the same _ slope or index _
@xmath10 for @xmath11 in many ob associations and star clusters in the milky way ( mw ) , the large- and small - magellanic clouds ( lmc , smc , respectively ) .
it is therefore suggested to refer to @xmath12 as the _ salpeter / massey slope _ or _ index_. it is valid for @xmath13 .
the imf is , strictly speaking , a hypothetical construct because any _ observed _ system of @xmath0 stars merely constitutes a particular _ representation _ of a distribution function .
the probable existence of a unique @xmath14 can be inferred from observations of many ensembles of such @xmath0 systems ( e.g. massey 2003 ) . if , after corrections for \(a ) stellar evolution , \(b ) unknown multiple stellar systems and \(c ) stellar - dynamical biases , the individual distributions of stellar masses is similar _ within the statistical uncertainties _ , then we ( the community ) deduce that the hypothesis that the stellar mass distributions are not the same can be excluded .
that is , we make the case for a _ universal , standard _ or _ canonical _ _ stellar imf _ within the physical conditions probed by the relevant physical parameters ( metallicity , density , mass ) of the populations at hand .
this canonical imf is a two - part - power law , the only structure with confidence found so far being the change of index from the salpeter / massey value to a smaller one near @xmath15 : @xmath16 it has been fully corrected for unknown multiple stellar systems in the low - mass ( @xmath17 ) regime , while multiplicity corrections in the high - mass regime await to be done . the evidence for a universal upper mass cutoff near @xmath18 ( weidner & kroupa 2004 ; figer 2005 ; oey & clarke 2005 ; koen 2006 ) seems to be rather well established in populations with metallicities ranging from the lmc ( @xmath19 ) to the super - solar galactic centre ( @xmath20 ) such that the stellar mass function ( mf ) simply stops at that mass .
this mass needs to be understood theoretically ( see discussion in kroupa & weidner 2005 ) .
chabrier ( 2003 ) offers a log - normal form to meet the needs of the observational data thereby losing its advantage .
a reason why the author prefers to use the canonical form is mathematical simplicity and the ease with which parts of it can be changed without affecting other parts . ] which fits the canonical form quite well ( e.g. romano et al . 2005 ) . below the hydrogen - burning limit
, there is substantial evidence that the imf flattens further to @xmath21 ( kroupa 2001a ; kroupa 2002 ; chabrier 2003 ) .
therefore , the canonical imf most likely has a peak at @xmath22 .
brown dwarfs , however , comprise only a few % of the mass of a population and are therefore dynamically irrelevant .
note that the logarithmic form of the canonical imf , @xmath23 , which gives the number of stars in log@xmath24-intervals , also has a peak near @xmath22 .
however , the _ system _ imf ( of stellar companions per binary combined to system masses ) has a maximum in the mass range @xmath25 ( kroupa et al .
2003 ) .
the above form has been derived from detailed considerations of local star - counts thereby representing an _ average _ imf : for low - mass stars it is a mixture of stellar populations spanning a large range of ages ( @xmath26 gyr ) and metallicities ( [ fe / h]@xmath27 ) , while for the massive stars it constitutes a mixture of different metallicities ( [ fe / h]@xmath28 ) and star - forming conditions ( ob associations to very dense star - burst clusters : r136 in the lmc ) .
therefore it can be taken to be a canonical form , and the aim is to test whether even more extreme star - forming conditions such as found in super - metal rich environments or super - dense regions may deviate from it
systematic _ deviations of the imf with physical conditions of the environment would constrain our understanding of star formation , would give us a prescription of how to set - up stellar - dynamical systems , and last not least would allow more precise galaxy - formation and evolution calculations .
there are two basic arguments suggesting that the imf ought to be dependent on the physical conditions of star formation : \a ) a region of a molecular cloud undergoing gravitational collapse will have over - dense sub - regions within it which are also jeans - unstable collapsing independently to form smaller structures which themselves may again sub - fragment ( e.g. zinnecker 1984 ) .
ultimately stars result .
the essence of this concept is that a region spanning a jeans length which has at least a jeans mass undergoes gravitational collapse .
the jeans mass depends on the temperature and density of the cloud , @xmath29 ( e.g. larson 1998 ; bonnell , larson & zinnecker 2006 ) .
now , in metal - rich environments there is more dust and therefore the collapsing gas can cool more effectively reducing @xmath30 and increasing @xmath31 .
thus , @xmath32 \ ; \uparrow \quad \longrightarrow \quad { \rm fragment\;masses}\ ; \downarrow\;.\ ] ] \b ) the fact that the imf is not a featureless power - law but has structure in the mass range @xmath33 suggests there to be a characteristic mass of a few @xmath34 . bonnell et al . ( 2006 ) suggest that this characteristic mass of fragmentation may be a result of the coupling of gas to dust such that there is a change from a cooling equation of state , where @xmath35 while the density increases , to one with slight heating at high densities , @xmath36 .
again , this implies a dependency of the characteristic mass on the metallicity through the cooling rate : @xmath32 \ ; \uparrow
\quad \longrightarrow \quad { \rm fragment\;masses}\ ; \downarrow\;.\ ] ] there seems to be observational evidence supporting the notion that stellar masses are derived from jeans - unstable mass fragments : pre - stellar cloud cores are found to be distributed like the canonical imf in low - mass star formating regions ( motte , andre & neri 1998 ; testi & sargent 1998 ; motte et al .
2001 ; but see nutter & ward - thompson 2006 ) .
while the concept of the jeans mass is very natural and allows one to nicely visualise the physical process of fragmentation , it has the problem that the densest regions of a pre - star - cluster cloud core ought to have the smaller fragment masses , but instead the most massive stars are seen to form in the densest regions .
a rather convincing physical model of the imf which avoids the problem with the jeans mass argument has been suggested by adams & fatuzzo ( 1996 ) and adams & laughlin ( 1996 ) .
the argument here is that the jeans mass has virtually nothing to do with the final mass of a star because structure in a molecular cloud exists on all scales .
therefore , no characteristic density can be identified , and `` no single jeans mass exists '' .
when a cloud region becomes unstable a hydrostatic core forms after the initial free - collapse .
this core then accretes at a rate dictated by the physical conditions in the cloud .
the metallicity influences the accretion rate through the sound velocity ( higher sound velocity , larger accretion rate ) by steering the cooling rate ( more metals @xmath37 more dust @xmath37 lower @xmath30 @xmath37 smaller sound velocity ) .
the many physical variables describing the formation of a single star have distributions , and folding these together yields finally an imf in broad agreement with a log - normal shape ( as also shown by zinnecker 1984 ) .
this fails though at large masses , where the imf is a power - law and where additional physical processes probably play a role ( coagulation , competitive accretion ) .
the important point however is that this theory _ also _ expects a variation of the resulting characteristic mass with metallicity as above : @xmath32 \ ; \uparrow \quad
\longrightarrow \quad { \rm characteristic\;mass}\ ; \downarrow\;.\ ] ] thus , both theoretical lines of argument seem to suggest the same qualitative behaviour , namely that the imf ought to shift to smaller characteristic masses with increasing metallicity .
this would therefore seem to suggest a _ very robust _ if not _ fundamental _ expectation of star - formation theory .
is it born out by empirical evidence ?
the best way to test this theoretical expectation is to measure the imf in metal - poor environments and to compare to the shape seen in metal - rich environments
. a measurement of the imf in a population with super - solar abundance would be especially important .
unfortunately this is extremely difficult , because populations with super - solar abundances are exceedingly rare .
only one star - cluster with [ fe / h]@xmath38 is known , ngc 6791 , but its ancient age of @xmath39 gyr implies it to be dynamically very old .
evaporation of low - mass stars will thus have significantly affected the shape of the present - day mass function , which has not been measured yet ( king et al . 2005 ) .
other metal - rich environments constitute the central mw region ( [ sec_kroupa : galc ] ) and galactic spheroids ( [ sec_kroupa : bulge ] ) , the latter allowing only indirect evidence on the imf through their chemical properties .
in general , measurements of the imf are hard , because stellar masses can only be inferred indirectly through their luminosity , @xmath40 , which also depends on age , metallicity and the star s spin angular momentum vector , collectively producing a distribution of @xmath40 for a given @xmath2 .
the empirical knowledge that can be gained about the imf is discussed next .
the best knowledge about the imf we glean from the local volume - limited stellar sample . the basic technique is to construct the stellar luminosity function ( lf ) , @xmath41 , where @xmath42 is the stellar absolute magnitude in some photometric pass band ( we are stuck with magnitudes rather than working with the physically more intuitive luminosities due to our historical inheritance ) .
the number of stars within the complete volume in the magnitude interval @xmath43 is then @xmath44 .
these are the same stars as enter @xmath45 from above , and thus results our master equation @xmath46 the observable is @xmath47 which we get from the sky .
our target is @xmath48 , and the hurdle is the _ derivative of the stellar mass luminosity relation _ , @xmath49 .
this is quite problematical , because we can only get at @xmath49 by either constructing observational mass luminosity relations using extremely well - observed binary stars with known kepler solutions which nevertheless have uncertainties that are magnified when considering the derivative , or we can resort to theoretical stellar models which give us well defined derivatives but depend on theoretically difficult processes within stellar interiors ( opacities , convection , rotation , magnetic fields , equation of state , nuclear energy generation processes ) .
there are two basic local lf s : ( i ) we can count all stars within a trigonometrically - defined distance limit such that the stellar sample is complete , i.e. we can see _ all _ stars of magnitude @xmath50 within a distance @xmath51 .
the volume - limited sample for solar - type stars having excellent parallax measurements extends to @xmath52 pc , while for the faintest m - dwarfs @xmath53 pc ( reid , gizis & hawley 2002 ) .
tests of completeness are made by comparing the stars with @xmath42 within @xmath51 to the number of these stars in a volume element further out ( henry et al . 2002 ) , finding that faint stars remain to be discovered even within a distance of 5 pc .
for this _ nearby lf _ , @xmath54 ,
the stars are well scrutinised on an individual - object basis , and geometric distances are known to within about 10 per cent . at the faint end @xmath55
is badly constrained resting on only a few stars .
\(ii ) prompted by the `` discovery '' of large amounts of dark matter in the mw disk ( bahcall 1984 ) , novel deep surveys were pioneered by reid & gilmore ( 1982 ) .
this second - type of sampling can be obtained by performing deep , pencil - beam photographic or ccd imaging surveys through the galactic disk . from the @xmath56 images
the typically 100 or so main sequence stars need to be gleaned using automatic image- , colour and brightness recognition systems .
the distances of the stars are determined using the method of _ photometric parallax _ , which relies on estimating the absolute luminosity of a star from its colour , and then calculating its distance from the distance modulus .
the resulting flux - limited sample of stars has photometric - distance limits within which the counts are complete .
the distance limits decrease for fainter stars .
clearly , while only _ one nearby lf _ exists , _ many photometric lfs _ can be constructed for different fields of view .
each observation yields a few dozen to hundreds of stars , and so the overall sample size becomes very significant .
the various surveys have shown @xmath57 to be invariant with direction .
this should , of course , be the case , since the galactic - field stars with an average age of about 5 gyr have a velocity dispersion of about 25 - 50 pc / myr such that within 200 myr a volume with a dimension of the survey volumes ( a few hundred pc ) is completely mixed .
it therefore came as a surprise that @xmath58 and @xmath59 are significantly different at faint luminosities ( fig .
[ fig_kroupa : mwlf ] ) .
-25 mm when not understanding something the best strategy to continue is sometimes to simply `` forget '' the problem and continue the path of least resistance .
thus , while the problem @xmath60 could not be explained immediately , it turned out to be constructive to first ascertain which lf shape must be the correct one using entirely different arguments . in eq .
[ eq_kroupa : lf_mf ] the slope of the stellar mass luminosity relation of stars enters posing a clue . fig .
2 in kroupa ( 2002 ) shows the mass luminosity data of binary stars with kepler orbits , and demonstrates that a non - linearity exists near @xmath61 such that a pronounced peak in @xmath62 appears at @xmath63 with an amplitude and width essentially identical to the maximum seen in the photometric lf at this luminosity ( fig .
[ fig_kroupa : mwlf ] ) .
this agreement of 1 .
the location of the maximum _ and _ 2 . the amplitude _ and _ 3 .
the width of the extremum convincingly suggest stellar astrophysics to be the origin of the peak in the lf , rather than the mf .
the wielen dip ( fig .
[ fig_kroupa : mwlf ] ) similarly results from subtle structure in the mass luminosity relation .
_ thus , simply by counting stars on the sky we are able to direct our gaze within their interiors _ : it is the internal constitution of stars which changes with changing main - sequence mass and this is what drives the structure in the mass luminosity relation .
having thus established that the peak in the lf _ must , in fact _ be there where it is also found as a result of fundamental astrophysical processes , we can test this result using star - clusters which constitute single - age , single - metallicity and equal - distance stellar samples . fig .
[ fig_kroupa : cllf ] does exactly this , and indeed , a very pronounced peak is evident at exactly the right location and with the right width and height ( kroupa 2002 : fig . 1 ) .
we can therefore trust the peak in @xmath57 . in fig .
[ fig_kroupa : mwlf ] it can be seen that @xmath64 also shows evidence for this peak , noting that the peak is smeared apart in the local stellar sample because the stars have a wide spread in metallicities .
the metallicity - dependence of the peak has been shown to be in agreement with the lfs of star clusters over a large range of metallicity ( von hippel et al .
1996 ; kroupa & tout 1997 ) .
kroupa , tout & gilmore ( 1993 ) then performed a trick to get at the correct mass luminosity relation without having to resort to theoretical or purely empirical relations which are very uncertain in their derivatives ( kroupa 2002 : fig .
2 ) : the malmquist - corrected @xmath57 is used to define the amplitude and width and location of the extremum in the single - age , single - metallicity average @xmath65 , and integrating the resulting constraint leads to a semi - empirical @xmath66 relation .
it is a semi - empirical relation because we used theoretical stellar models to place , one can say , zeroth order constraints on this relation , i.e. to prove the existence of the extremum and to estimate its location and width and amplitude . with this theoretical knowledge in - hand
we then used the lf to constrain the detailed run of @xmath66 .
the result is in amasing agreement even with the most recent high - quality binary - star constraints by delfosse et al .
( 2000 ) . with the so - obtained @xmath66 relation , which has the correct derivative , it is now possible to take another step towards constraining the mf . but first the problem @xmath67 needs to be addressed .
one bias clearly affecting @xmath57 are unresolved multiple systems ; in constructing @xmath64 all stellar companions have been counted individually , while @xmath57 consists of `` stars '' too far away to be resolved if they are multiple .
also , a faint companion star may not be seen because it lies below the flux limit while the primary enters the formal photometric distance limit . furthermore , an unresolved binary system appears redder unless both stars are of equal mass and this affects the calculated stellar space density because the photometric distance is misjudged .
all the effects can and must be corrected for ; this having been done thoroughly for the first time by kroupa et al .
( 1993 ) . the bias due to unresolved binaries affecting @xmath57 can be nicely demonstrated using realistic numbers for the multiplicity in the galactic field ( goodwin et al .
2006 ) : suppose the observer sees 100 systems on the sky .
of these 40 are binaries , 15 triples and 5 quadruples .
the multiplicity fraction is @xmath68 .
the observer will construct the lf using these 100 systems , but in fact 85 stars are missed .
all of these are much fainter than their primary , implying a non - linear depression at the faint end of the lf . together with a thorough modelling of the star - formation history of the solar neighbourhood ( low - mass stars take long to descend to the main sequence ) , local galactic disk structure and metallicity spreads , as well as different fractions of multiple stars and photometric and trigonometric distance uncertainties , kroupa et al .
( 1993 ) performed a multi - dimensional optimisation for the parameters of a two - part - power law mf .
the result is one mf which unifies both @xmath64 and @xmath57 .
it is given by eq .
[ eq_kroupa : canonimf ] for @xmath69 ( kroupa et al .
1993 ; reid et al .
2002 ) . adopting this imf as being correct ,
kroupa ( 1995b ) also showed that the observed lfs are arrived at if stars are born as binaries in clusters which evolve and disperse their stellar content into the galactic field .
chabrier ( 2003 ) further points out that non - linearities in the colour magnitude relation used in photometric parallax adds to the underestimate of the stellar densities at the faint end of @xmath57 .
scalo ( 1986 ) had performed a very detailed analysis of the stellar mass function , which was superseded for @xmath69 through new star - count data and the new modelling described above . but for @xmath70 his analysis remains valid .
using star counts out to kpc distances for early - type stars he modelled their spatial distribution and took account of their production through different star - formation rates , and arrived at an estimate of the galactic - field imf adopted in kroupa et al .
( 1993 ) : @xmath71 together with eq .
[ eq_kroupa : canonimf ] this is the _ standard galactic - field imf_. elmegreen & scalo ( 2006 ) then showed that there can be artificial features in the massive - part of the galactic - field imf when it is deduced from the present - day field mf under varying star - formation rates ( sfrs ) , frustrating attempts to attribute any possible structure there to star - formation physics . as already stated in
[ sec_kroupa : intro ] , the imf for early - type stars can also be inferred for individual ob associations and star clusters .
massey ( 2003 and previous papers ) has shown the imf for massive stars to be independent of density ( @xmath72 stars / pc@xmath73 , the densest cluster being r136 in the lmc ) and metallicity ( @xmath74 , the population with the lowest metal abundance being in the smc ) .
it always has @xmath75 , thus leading to the standard or canonical stellar imf , eq .
[ eq_kroupa : canonimf ] .
the question now emerges as to why @xmath76 for @xmath77 .
this is referred to as the `` scaslo _ vs _ massey discrepancy '' .
considering the uncertainties , it would be valid to discard the difference .
however , using the standard galactic - field imf in galactic evolution models would yield about 3 times fewer stars in the mass range @xmath78 and about 8 times fewer stars with @xmath18 than when using the canonical imf .
so it is important to know @xmath79 .
we will return to this later - on , but as before , when not immediately seeing a possible solution it often proves useful to ignore the problem until new insights open new avenues for exploration .
spectroscopic and speckle - interferometric observations of nearby massive star - forming regions have shown that ob stars prefer more than one partner ( zinnecker 2003 ; goodwin et al .
in particular , preibisch et al .
( 1999 ) present the most detailed multiplicity study of the orion nebula cluster ( onc ) demonstrating that the rule are triples rather than binaries as amongst t tauri stars .
this allows us to set - up the following _ hypothesis _ : @xmath80 that is , massey was observing dynamically less - evolved and therefore more binary - rich populations whereas scalo saw the field population which is , to a significant extend , `` contaminated '' by runaway ob stars which are mostly single ( stone 1991 ) .
this would be a neat solution to our problem : fainter ob stars are hidden among their brighter primaries flattening the massey - imf , while the mostly single ob - field stars in the scalo sample would be a more correct representation of the true imf . thus we may be led to conclude that in actuality the true ( binary - corrected ) imf is eq .
[ eq_kroupa : scaloimf ] ( @xmath81 ) , while eq .
[ eq_kroupa : canonimf ] ( @xmath82 ) is wrong .
this would , of course , have important implications for galactic astrophysics as well as cosmology ; most galaxy - formation and evolution calculations are being done with a salpeter imf .
star clusters pose star - formation events which are extremely well correlated in space , time and chemistry , and so studying the mf within clusters avoids most of the problems we need to deal with in the galactic field .
however , there are a number of serious issues that essentially annihilate any advantage one might have gained by having an equal - age , -distance and -metallicity population : to avoid dynamical evolution complicating the star - counts in a cluster , young clusters are chosen .
however , uncertainties in pre - main sequence tracks make the calculation of stellar masses from stellar luminosities and colours or spectra very uncertain for ages less than a few myr . errors of about 50 % would not be untypical .
these errors are not randomly distributed though , but cause systematic offsets from the true stars . and so unphysical structures appear in the mf .
this is discussed in more depth in kroupa ( 2002 ) .
very young clusters also have a high binary fraction , which again would require corresponding corrections .
young clusters evolve very rapidly during the first 1 myr as a result of residual gas expulsion .
for example , the onc with an age of about 1 myr is already between 5 and 15 initial crossing times old , implying a rapid dynamical evolution of the binary population ( fig . 1 in kroupa 2000 ; fig .
9 in kroupa et al .
2001 ) _ and _ an expansion of the cluster such that a substantial fraction of its initial population has probably been lost ( kroupa , aarseth & hurley 2001 ) .
this translates into a selective stellar - mass loss _ if _ the cluster was initially significantly mass - segregated ( moraux , kroupa & bouvier 2004 ) .
it is therefore never clear exactly which corrections have to be applied , and an inferred stellar mf is very likely not to be a good representation of the imf . to avoid the issues above ( uncertain pre - main sequence theory , rapidly evolving stellar and binary populations )
, older clusters within which stars are on or close to the main sequence would appear to be useful .
they are definitely useful , but such clusters are heavily evolved dynamically , and so the mf will not represent the imf even for clusters with an age near 100 myr .
unresolved multiple systems remain a problem , and for example for the pleiades stauffer ( 1984 ) shows that the fraction of photometric binaries is 26 % ( these are stars that are more luminous than a single star of the same colour ) , while kaehler ( 1999 ) notes that a true binary fraction between 60 and 70 % may be possible .
even globular clusters appear to have a sizeable binary population ( e.g. rubenstein & bailyn 1997 ) . a star cluster which has managed to survive its initial gas expulsion phase re - virialises on a time - scale of tens of myr ( kroupa et al . 2001 ) .
it s binary population is depleted by then such that binaries with orbital kepler velocities smaller than the velocity dispersion in the pre - expansion cluster will have been disrupted .
the remaining binaries are mostly inert ; sometimes energetic binary single - star or binary binary encounters near the cluster core eject 13 stars ( e.g. a single star and the binary ) from the cluster .
the cluster s evolution is , however , well described by classical dynamical evolution tracks , i.e. the surviving binaries are dynamically not important , as shown explicitly by kroupa ( 1995c ) . during this long - lived phase the cluster evolves towards energy equipartition , which implies that the low - mass stars gain energy and
thus move outwards becoming lost to the galactic tide , while the more massive stars loose energy and segregate towards the center of the potential .
this process never stops , i.e. a cluster never actually reaches energy equipartition and dynamical equilibrium .
the consequence is that as the cluster ages , it s mf becomes increasingly depleted in low - mass stars .
this is nicely evident in low - mass clusters in terms of a flattening of the lf , while unresolved binaries remain a significant bias despite their dynamical unimportance ( fig .
10 in kroupa 1995c ) .
baumgardt & makino ( 2003 ) study the evolution of the mf in much detail for massive star clusters without binaries . in their fig .
7 they nicely show how the mf changes its slope in dependence of the cluster s dynamical age expressed in terms of the fraction of the time until cluster disruption : clusters that have passed 50 % of their disruption time have essentially a flat ( @xmath83 ) mf for @xmath84 , while for older clusters @xmath85 becomes negative .
two processes compete for all stellar masses , but are particularly pronounced for massive stars because of the involved energetics .
* mass segregation : * if massive stars form throughout a forming star cluster energy equipartition will force them to segregate to the center within a timescale @xmath86 , where @xmath87 are the masses of the average star and the massive stars , and @xmath88 is the median two - body relaxation time , as detailed in kroupa ( 2004 ) . for example , for the onc , @xmath89 myr and @xmath90 such that @xmath91 myr @xmath92 age of the onc . no wonder that the onc sports a beautiful trapezium , although it may also have formed at the centre ( bonnell & davies 1998 ) .
* core decay : * the massive stars at the cluster centre , which may have been born there or segregated there , form a short - lived small-@xmath93 core .
it decays on a time - scale @xmath94 , where @xmath93 is the number of massive stars in the core and @xmath95 is its crossing time .
again , for the onc , we find ( kroupa 2004 ) @xmath96 yr @xmath92 its age .
so why does the trapezium still exist ?
detailed @xmath0-body computations confirm this time - scale problem ( pflamm - altenburg & kroupa 2006 ) : the onc also has a highly abnormal `` imf '' - only 10 stars more massive than @xmath97 are found in it , while 40 would be needed to allow the very final remnant of the central core to be still visible today _ and _ to account for the mass of the most massive star , given the cluster - mass maximal - mass relation observed to exist among young clusters ( weidner & kroupa 2006 ) .
this appears to be the rule rather than the exception , pflamm - altenburg & kroupa ( 2006 ) discuss the older upper scorpius ob association which also shows a deficit of massive stars compared to the canonical imf .
_ naively this may be taken as good evidence for a bottom - heavy present - day imf in a relatively metal - rich environment _ , which would even be consistent qualitatively with the theoretical arguments of
[ sec_kroupa : theory ] ( higher - metallicity environments making lower - mass stars on average )
. further work , however , shows this to be an illusion : pflamm - altenburg & kroupa ( 2006 ) apply a specially designed chain - regularisation code ( catena ) to study the dynamical stability of onc - type cores , finding that if the onc and upper scorpius ob association , which is understood to be an evolved version of the onc , were born with 40 ob stars , then the observations are well - accounted for .
indeed , stone ( 1991 ) , among others , shows that 46 % of all o stars are probably runaways of which about 18 % have line - of - sight velocities @xmath98 pc / myr .
energetic dynamical encounters in cluster cores therefore have a very significant effect on the shape of the `` imf '' in clusters and ob associations . _ and
_ , massive stars are rapidly dispersed from their birth sites with corresponding implications for the spreading of heavy elements and feedback - energy deposition throughout a galaxy .
the implications of these studies is therefore that the massive - star content of star clusters and ob associations is not a measure of the true initial content , unless the full dynamical history of the cluster or ob association is taken into account .
it would thus appear that clusters and ob associations of any age are a _
horrible _ place to study the imf .
this does not mean , however , that these systems are useless .
quite on the contrary : they are indeed our only samples of stars formed together and are therefore the only samplings from the imf that exist before dispersion into the field .
the point of the above pessimistic note is to instill the fact that any measurement of the `` imf '' in a cluster or association _ must _ be accompanied by a dynamical investigation _ before _
conclusions about the possible variation of the imf can be made .
an excellent example follows in [ sec_kroupa : galc ] .
the derivation of mfs in many clusters and associations leads to many values of @xmath85 measured over different stellar mass ranges .
the alpha - plot , @xmath99 , can therefore give us information about the true shape of the putative underlying imf , and also of the scatter about a mean imf ( scalo 1998 ) .
this scatter can then be studied theoretically using the methods mentioned above , i.e. @xmath0-body calculations of star - clusters with different properties .
thus , [ sec_kroupa : pessimism ] needs to be invoked in the context of some recent interesting observational studies that suggest the imf to vary at low masses ( e.g. oasa et al .
2006 ) and at high masses ( gouliermis et al .
2005 ) .
one important aspect here is the purely statistical scatter resulting from finite-@xmath0 sampling from a putative universal imf ( [ sec_kroupa : intro ] ) .
elmegreen ( 1997 , 1999 ) showed that the observed scatter is consistent with the variations of @xmath99 as a result of sampling from the imf .
stellar - dynamical and unresolved binary - star biases lead to systematic deviations of @xmath99 away from the true value , and kroupa ( 2001 , 2002 ) has shown the observed alpha - values to be consistent with the canonical imf ( eq .
[ eq_kroupa : canonimf ] ) for a large variety of stellar populations using theoretical models of initially binary - rich star clusters .
as already deduced with some force by massey ( 2003 , and previous papers ) , there is no evidence for systematic shifts of @xmath99 with metallicity nor density . at the one extreme end of the scale of physical conditions , globular clusters have very similar mfs as young galactic clusters and the galactic field ,
while constraints on the massive - star content of globular clusters are such that the imf could not have been too top heavy as otherwise the mass - loss from evolving stars would have unbound the clusters ( kroupa 2001 ) . at the other extreme end , the super - solar arches cluster near the galactic centre has a massive - star imf again very similar to the salpeter / massey one ( [ sec_kroupa : galc ] below ) .
kroupa ( 2002 ) studied the distribution , @xmath100 , of measured @xmath85 values for stars more massive than a few @xmath101 finding it to be describable by two gaussians , one somewhat broad but with symmetric wings , and the other exceedingly narrow , and both centered on the salpeter / massey value @xmath102 .
this is very surprising because the various @xmath99 values were obtained by different research groups observing different objects .
an equivalent theoretical data set shows a somewhat broader unsymmetric distribution offset from the input salpeter / massey value .
this needs further investigation , because one would expect the theoretical data to be better `` behaved '' than the observational ones which include the entire complications of inferring masses from observed luminosities . concerning this open question , maz apellniz & beda ( 2005 ) point out that binning the star - counts into mass - bins of equal mass leads to biases that distort the shape of the imf when the number of stars is small , which is often the case at high masses .
returning to the empirically - determined existence of an upper mass limit to stars near @xmath18 ( [ sec_kroupa : intro ] ) , it remains to be understood why it does not seem to depend on metallicity : theoretical concepts would suggest that feedback ought to play a role in limiting the masses of stars , but this implicitly contains a metallicity dependence as the coupling of radiation to the accreting gas is more efficient in metal - rich environments .
thus , we are left yet again with two open questions .
one important implication from this analysis of @xmath99 values for massive stars is that the above hypothesis ( eq . [ eq_kroupa : hyp ] ) must be discarded : if unresolved multiple systems were the origin of the _ scalo vs massey
_ discrepancy , then @xmath100 would have an asymmetric distribution stretching from the single - star value ( @xmath103 ) to the unresolved binary - star value ( @xmath104 ) .
this is not the case .
the scalo vs. massey discrepancy therefore is either non - existing ( perhaps because scalo s analysis requires an update ) , or it has another hitherto unknown origin .
the proposition by elmegreen & scalo ( 2006 ) that the imf calculated from the present - day field mf may be steeper than the canonical imf if the field was created during a lull in the star - formation rate deserves some attention in this context .
reports do exist that in extreme environments there is evidence for a top - heavy imf .
these often come from indirect arguments based on luminosity and available mass for star formation in star - bursts , for example : scaling the canonical imf to the observed luminosity would lead to a mass in stars far superior to dynamical mass constraints from kinematics , requiring either a top - heavy ( smaller @xmath79 ) imf , or an imf truncated typically near or above @xmath105 .
paumard et al .
( 2006 ) report evidence for a top - heavy imf in both circum - nuclear disks in the mw .
while the observations are quite convincing , [ sec_kroupa : pessimism ] needs to be invoked before taking such results as affirmative evidence for a variable stellar imf .
a case in point is the arches cluster at the galactic centre , where the conditions for star formation are very different to those prevalent in the solar vicinity or the disk of the lmc , by being in general warmer ( @xmath98 k instead of 10 k ) and having a super - solar metallicity ( @xmath106[fe / h]@xmath107 ) .
klessen , spaans , jappsen ( 2006 ) perform an excellent state - of - the art calculations of star - formation under such conditions .
they state in their abstract : `` in the solar neighbourhood , the mass distribution of stars follows a seemingly universal pattern . in the centre of the milky way
, however , there are indications for strong deviations and the same may be true for the nuclei of distant star - burst galaxies . here
we present the first numerical hydro - dynamical calculations of stars formed in a molecular region with chemical and thermodynamic properties similar to those of warm and dusty circum - nuclear star - burst regions .
the resulting initial mass function is top - heavy with a peak at @xmath108 , a sharp turn - down below @xmath109 and a power - law decline at high masses .
we find a natural explanation for our results in terms of the temperature dependence of the jeans mass , with collapse occurring at a temperature of @xmath110 k and an h@xmath111 density of a few @xmath56 @xmath112 , ... '' .
theirs would be in agreement with the previously reported apparent decline of the stellar mf in the arches cluster below about @xmath113 . shortly after the above theoretical study appeared , kim et al .
( 2006 ) published their observations of the arches cluster and performed the necessary state - of - the art @xmath0-body calculations of the dynamical evolution of this young cluster , revising our knowledge significantly . quoting from their abstract : `` we find that the previously reported turnover at @xmath113 is simply due to a local bump in the mass function ( mf ) , and that the mf continues to increase down to our 50 % completeness limit ( @xmath114 ) with a power - law exponent of @xmath115 for the mass range of @xmath116 .
our numerical calculations for the evolution of the arches cluster show that the @xmath117 values for our annulus increase by 0.1 - 0.2 during the lifetime of the cluster , and thus suggest that the arches cluster initially had a gamma of @xmath118 @xmath119 , which is only slightly shallower than the salpeter value . ''
( @xmath120 ) .
this serves to illustrate the general observation that when detailed high - resolution observations _ and _ accurate @xmath0-body calculations are combined , claims of abnormal imfs tend to vanish , irrespective of what theory prefers to have .
another example , the onc , has already been noted above .
an interesting , albeit indirect , result by francesca matteucci and her collaborators based on multi - zone photo - chemical evolution studies of the mw , its bulge and elliptical galaxies shows the mw disk to be re - producible by the standard galactic - field imf ( eq .
[ eq_kroupa : scaloimf ] , @xmath81 ) but not the canonical imf ( eq .
[ eq_kroupa : canonimf ] ) .
this is supported by the independent work of portinari , sommer - larsen & tantalo ( 2004 ) .
in contrast , the mw bulge may require a rapid phase of formation lasting 0.010.1 gyr and a top - heavy imf with @xmath121 ( romano et al .
2005 ; ballero , matteucci & origlia 2006 ) . for elliptical galaxies
a similar scenario seems to hold , whereby the imf needed to account for the colours and chemical properties is the canonical one ( @xmath122 ) , but flattening ( decreasing @xmath79 ) slightly with increasing galaxy mass ( pipino & matteucci 2004 ) .
an experiment , where the yields for massive stars are changed as much as possible without violating available constraints confirms this result - it seems not to be possible to understand the metallicity distribution of mw bulge stars with a standard galactic field imf .
instead , @xmath123 seems to be required ( ballero , kroupa & matteucci 2007 ) .
nevertheless , uncertainties remain .
thus , samland , hensler & theis ( 1997 ) find the bulge to have formed over a time - span of about 4 gyr with a salpeter imf using a two - dimensional chemo - dynamical code , while immeli , samland & gerhard ( 2004 ) obtain formation time - scales near 1 gyr using a three - dimensional chemo - dynamical code .
finally , zoccali et al .
( 2006 ) find the `` mw bulge to be similar to early - type galaxies , in being @xmath85-element enhanced , dominated by old stellar populations , and having formed on a timescale shorter than @xmath1241 gyr '' .
`` therefore , like early - type galaxies the mw bulge is likely to have formed through a short series of starbursts triggered bythe coalescence of gas - rich mergers , when the universe was only a few gyr old '' . for low - mass stars , zoccali et al .
( 2000 ) find the bulge to have a similar mf as the disk and globular clusters .
the `` scalo _ vs _ massey discrepancy '' found in [ sec_kroupa : canimf ] and its proposed solution in
[ sec_kroupa : massive_bins ] in terms of unresolved multiple systems was found , in
[ sec_kroupa : sc_mas ] , not to be possible .
again , as in [ sec_kroupa : canimf ] we might argue that trying to solve the discrepancy is not worth the effort , given the uncertainties .
indeed , the probable solution drops - out through an entirely different line - of - thought : it needs to be remembered that star clusters are the _ fundamental building blocks of galaxies _ ( kroupa 2005 ) .
therefore , to be exact , all the stellar imfs in all the clusters forming in one `` star - formation epoch '' ( weidner , kroupa & larsen 2004 ) of a galaxy must be added to calculate the global , galaxy - wide `` integrated galactic imf '' ( igimf ) . this has been pointed - out by vanbeveren ( 1982 , 1984 ) and discussed again by scalo ( 1986 ) , and formulated for an entire galaxy by kroupa & weidner ( 2003 , 2005 ) , noting that the star - clusters are distributed as a power - law embedded - cluster mass function with an exponent @xmath125 , where the salpeter value would be 2.35 .
also , an empirical relation exists limiting the mass of the most massive star in a cluster with stellar mass @xmath126 : @xmath127 ( weidner & kroupa 2006 ) , enhancing the _ igimf@xmath128imf effect_. putting this together , an integral over all clusters formed together in one epoch results , and the composite imf turns out to be steeper with @xmath129 ( @xmath130the canonical imf value ) for @xmath131 with a steep downturn at a mass @xmath132 .
both , @xmath133 and the galaxy - wide maximum star mass , @xmath134 , depend on the star - formation rate of the galaxy , and so the situation becomes complex with important implications for galactic spectro - photometric and chemical evolution studies given that the number of supernovae of type ii is depressed significantly per unit stellar mass formed when compared to a universal salpeter imf , which is often assumed in cosmological applications ( goodwin & pagel 2004 ; weidner & kroupa 2005 ) .
koeppen , weidner & kroupa ( 2006 ) show that this igimf - ansatz naturally explains the observed form of the mass metallicity relation of galaxies without the need to invoke selective outflows ( which are not excluded though ) .
applied to the mw , scalo s @xmath81 comes - out naturally for an input canonical imf .
this is also required by spectro - photometric and chemical - evolution models of the mw disk ( [ sec_kroupa : bulge ] ) . at the same time , for lsb galaxies the igimf must be steep , i.e. bottom - heavy .
again , this seems to be confirmed by observations ( lee et al .
2004 ) . is this , then , the solution to the `` scalo _ vs _ massey discrepancy '' ? elmegreen ( 2006 ) , however , opposes the notion that the igimf is different to the stellar imfs found in individual clusters , mostly because the star - cluster mass function is not steep steep enough ( @xmath135 ) , in which case @xmath136 as already shown by kroupa & weidner ( 2003 ) . _
much therefore rests on the detailed shape of the star - cluster initial mass function _ !
also , the existence of a physical @xmath137 relation is of central importance to the @xmath138 argument , but opposing views are being voiced ( e.g. elmegreen 2006 ) .
an argument against igimf @xmath139imf is that some studies suggest these to be equal , e.g. in elliptical galaxies ( [ sec_kroupa : compimf ] ) .
however , an appreciable fraction of the research which elmegreen fields in - support of his equality - conjecture actually support the inequality - conjecture .
so this issue is still very much under debate and may need observational improvement by ( i ) better constraining the true initial star - cluster mass function and ( ii ) improving the empirical @xmath140 relation .
observational work is underway to test the igimf@xmath128imf notion , and for example selman & melnick ( 2005 ) find the composite imf to be flatter than the stellar imf in the 30 dor star - forming region , as predicted , but only with low - significance .
it would have seemed that when two mutually more or less exclusive theories of the origin of stellar masses make the same basic prediction concerning the variation of the average stellar mass with physical conditions , that this expected variation would be very robust and born out in observational data .
alas , the observational data on the imf are resilient - they do not yield what we desire to see .
the stellar imf is invariant and can be best described by a two - part power - law form ( eq . [ eq_kroupa : canonimf ] ) .
this holds true for metallicities ranging from those of globular clusters to super - solar values near the galactic centre , and for densities less than about @xmath141 stars / pc@xmath73 .
even the maximal stellar mass of about @xmath18 seems to be independent of metallicity for @xmath142 .
given the quite total failure to account for the resilience of the stellar imf towards changes , it is clear that this imf - conservatism poses some rather severe challenges on liberal star formation theory .
only indirect arguments based on chemical evolution work seem to suggest the imf to have been somewhat flatter for massive stars during the extreme physical conditions valid during the formation of the mw bulge and elliptical galaxies .
variations of the galaxy - wide imf ( the igimf ) among galaxies and also in time can be understood today despite the existence of a universal canonical imf , but this issue is still very much in debate ( elmegreen 2006 _ vs _ weidner & kroupa 2006 ) .
the evidence noted in [ sec_kroupa : bulge ] appears to suggest that perhaps @xmath143 for the mw disk , @xmath144 for e - galaxies , while @xmath145 for massive e - galaxies and the mw bulge .
is this true ? :
i would like to thank the organisers sincerely for a most enjoyable meeting and my collaborators for important contributions .
this research was supported by an isaac newton and a senior rouse ball studentship in cambridge in the uk , and by the heisenberg programme of the dfg and dfg research grants in germany .
this article was mostly written in vienna and i thank christian theis and gerhardt hensler for their hospitality .
|
does the imf vary ? is it significantly different in metal - rich environments than in metal - poor ones ?
theoretical work predicts this to be the case .
but in order to provide robust empirical evidence for this , the researcher _ must _ understand _ all _ possible biases affecting the derivation of the stellar mass function .
apart from the very difficult observational challenges , this turns out to be highly non - trivial relying on an exact understanding of how _ stars evolve _ , and how stellar populations in galaxies are assembled dynamically and how individual star clusters and associations evolve .
_ @xmath0-body modelling _ is therefore an unavoidable tool in this game : the case can be made that without complete dynamical modelling of star clusters and associations any statements about the variation of the imf with physical conditions are most probably wrong . the calculations that do exist demonstrate time and again that the imf is invariant : there exists no statistically meaningful evidence for a variation of the imf from metal - poor to metal - rich populations .
this means that currently existing star - formation theory fails to describe the stellar outcome .
indirect evidence , based on chemical evolution calculations , however indicate that extreme star - bursts that assembled bulges and elliptical galaxies may have had a top - heavy imf .
[ 1996/06/01 ]
|
results from the wilkinson microwave anisotropy probe ( wmap ) , especially the first - year data @xcite and to some extent the three - year data @xcite , have placed a focus on possible running of the spectral index of density perturbations ( see e.g. lidsey & tavakol 2003 ; kawasaki , yamaguchi & yokoyama 2003 ; chung , shiu & trodden 2003 ; bastero - gil , freese & mersini - houghton 2003 ; chen et al . 2004 ; covi et al . 2004 ; ashoorioon , hovdebo & mann 2005 ; ballesteros , casas & espinosa 2006 ; cline & hoi 2006 ; corts & liddle 2006 ; easther & peiris 2006 ) .
it is certainly premature to draw any strong conclusions based on existing evidence , especially as it remains controversial whether current data even support power - law models over the harrison
zeldovich ( hz ) model , but it is timely to investigate the extent to which the upcoming planck satellite may resolve the situation . as we have stressed in several recent papers ( e.g. mukherjee , parkinson & liddle 2006a ; parkinson , mukherjee & liddle 2006 ; liddle , mukherjee & parkinson 2006a ) , the appropriate statistical tool for assessing the need to introduce new parameters is _ model selection _ @xcite .
model selection assigns probabilities to _ sets _ of parameters , i.e. models , in addition to the usual probability distributions for parameter values within each model .
for example , bayesian model selection applied to data compilations including wmap3 shows that the case for including even just the spectral index @xmath1 as a variable fit parameter is inconclusive @xcite . in a recent paper @xcite
, we used model selection forecasting tools to assess the ability of the planck satellite to distinguish between the harrison zeldovich model with @xmath2 and a model with varying spectral index , vary@xmath3 .
the outcome naturally depends on the assumed true value of @xmath1 , which we call the _ fiducial value _ , and we found that planck can strongly favour the latter model only if the true value of @xmath1 lies outside the range @xmath4 $ ] . in making that comparison , we assumed that the true spectrum could be described by a power - law . in this paper
, we extend that analysis to include the possibility of spectral index running , given by .
this adds an extra model , vary@xmath5 , to the model set .
this means that we are carrying out a three - way model comparison , within the two - dimensional space defined by the fiducial values of @xmath1 and @xmath6 .
ideally we would also have included tensor perturbations in this analysis in order to fully represent the usual inflationary predictions ( e.g. liddle & lyth 2000 ) , but the present analysis is at the limits of current computer power , having required many months of multi - processor time .
our approach exactly follows our earlier paper @xcite , and so we provide only the briefest of summaries here and refer to that paper and references therein for details .
model selection forecasting was first introduced by trotta ( 2007b ) , whose predictive posterior odds distribution ( ppod ) forecasting determined the probability of different model selection outcomes of future experiments based on present knowledge .
an alternative approach , which delineates regions of parameter space where different model selection verdicts are expected , was introduced in mukherjee et al .
( 2006b ) ; a combination of the methods was used in pahud et al .
( 2006 ) , and also subsequently in trotta ( 2006 ) , liddle et al .
( 2006b ) and trotta ( 2007a ) .
as used here , data are simulated for the different possible true values of the parameters of interest , known as fiducial values , and a model comparison analysis carried out at each point .
although not required , in typical examples a simple model will be nested within a more complex one , e.g. the hz model is the special case of vary@xmath3 with @xmath2 . if the ( assumed ) true model is the nested one , the model comparison will favour that model , and one may ask how strongly . if instead the true model is the more complex one , one can ask how far from the simple model the true values would have to be , in order that a given experiment can overcome statistical uncertainty and deliver a strong or decisive verdict in favour of the complex model .
these two notions can be used to define model selection figures - of - merit , assessing the abilities of competing experiments @xcite . in our work ,
we use the bayesian evidence @xmath7 as the model selection statistic . like any model selection statistic
, it creates a tension between goodness of fit to the data and the complexity of the model .
it represents a full implementation of bayesian inference , being the probability of the data given the model ( i.e. the model likelihood ) .
it updates the prior model probability to the posterior model probability .
computations are carried out using the nested sampling algorithm @xcite , using our code cosmonest @xcite .
computing the evidence accurately is significantly more challenging than computing the posterior probability distribution , and so the calculations are computationally time - consuming .
our assumption is that there are three models of interest in fitting future planck data .
these are the harrison zeldovich model , a power - law model where @xmath1 is fit from data , and a model where both @xmath1 and @xmath6 are varied .
we denote these models hz , vary@xmath3 , and vary@xmath5 respectively , and also indicate them by use of subscripts 0 , 1 , and 2 respectively . in the presence of running
, the spectral index is defined in the usual way by @xmath8 the pivot scale @xmath9 corresponds to a scale well constrained by existing data . when running is included , @xmath10 is always specified at this scale , and throughout we assume the running is constant . as in pahud et al .
( 2006 ) , the prior range for @xmath1 is taken to be @xmath11 , representing a reasonable range allowed by slow - roll inflation models ( see e.g. liddle & lyth 2000 ) .
we take the prior on @xmath6 to be @xmath12 .
this is somewhat arbitrary .
slow - roll inflation models would tend to suggest a much smaller value @xcite , but there is no point in restricting the analysis to values smaller than planck can measure , as one will simply conclude that planck is unable to make the measurement . accordingly
, our range is loosely motivated by present observational knowledge , corresponding to models with unexpectedly large running .
the comparison between two models does have some prior dependence on the extra parameter(s ) .
if one prior is widened in regions where the likelihood is negligible , then the evidence changes proportional to the prior volume , so for instance a doubling of the prior range will only reduce @xmath13 by @xmath14 . in running cosmonest , the algorithm parameters used were @xmath15 live points and an enlargement factor of 1.8 for hz , 1.9 for vary@xmath3 , and 2.0 for vary@xmath5 .
the tolerance parameter was set to 20 ( rather than 0.5 as in our previous analysis ) in order to improve the speed of the simulations .
this is sufficient to give answers to good accuracy as indicated by the uncertainties obtained .
four independent evidence evaluations were done for each calculation , to obtain the mean and its standard error .
we then compare our models in pairs by considering the bayes factor , defined as the ratio of evidences between two models , written @xmath16 , for @xmath17 ( @xmath18 ) , where @xmath19 and @xmath20 indicate the two models under assumption . by plotting the bayes factor using datasets generated as a function of the two parameters of interest ,
one uncovers the regions of the two - dimensional fiducial parameter space in which the planck satellite would be able to decisively select between the two models , and also those regions where the comparison would be inconclusive . in order to assess the significance of any difference in evidence between two models , a useful guide is given by the jeffreys scale @xcite .
labelling as @xmath19 the model with the higher evidence , it rates as ` not worth more than a bare mention ' , @xmath21 as ` substantial ' , @xmath22 ` strong ' to ` very strong ' , and @xmath23 as ` decisive ' .
we simulate planck data exactly as described in pahud et al .
having determined the fiducial model power spectra , we simulate temperature power spectrum data for the three most sensitive high frequency instrument ( hfi ) channels and the polarization signal for only one of these channels , modelling instrument noise using current detector specifications .
the simulations are somewhat simplistic , as computational limitations prevent a more detailed treatment that might include residuals from foreground subtraction and @xmath24 noise .
however they should provide a good characterization of the planck data for our purposes .
simulations are carried out for various values of the spectral index and its running , and the other parameters are those of the usual @xmath25cdm model in a flat spatial geometry .
the logarithm of the bayes factor , @xmath26 , as a function of the fiducial value of @xmath1 .
the horizontal lines indicate where the comparison becomes ` strong ' ( dashed ) and ` decisive ' ( solid ) on the jeffreys scale.,width=283 ] in simulating the data , we are primarily interested in the dependence on the key parameters of interest , @xmath1 and @xmath6 , and different data simulations are carried out for a grid of values in that plane .
the other cosmological parameters are given fixed fiducial values as in pahud et al .
( 2006 ) , namely the baryon physical density @xmath27 , the cold dark matter physical density @xmath28 , the sound horizon @xmath29 , the optical depth @xmath30 , and the density perturbation amplitude normalization @xmath31 .
the corresponding value of the hubble parameter is @xmath32 .
the model selection verdict should have negligible dependence on these fiducial values .
note that all parameters , including these , are varied in computing the evidences of the models ; it is only in defining the fiducial models for data simulation that these parameters are fixed .
the prior ranges used for these parameters are as in pahud et al .
( 2006 ) : @xmath33 , @xmath34 , @xmath35 , @xmath36 , and @xmath37 .
we begin by showing in fig .
[ f : plot_lnb_01_1d ] the main result obtained in pahud et al .
( 2006 ) . in that analysis
, running was not included and so the fiducial @xmath6 is zero . at @xmath2 , corresponding to hz being the true model , the hz model is strongly preferred with @xmath38 .
it has a higher evidence since it can fit the data just as well as vary@xmath3 and has one less parameter .
once @xmath1 is far enough away from 1 , the hz fit becomes very poor and the bayes factor plummets .
the speed with which this happens indicates the strength of the experiment .
the vary@xmath3 model becomes strongly favoured only once @xmath39 or @xmath40 ; if the true value lies within that range even the planck satellite will give inconclusive results .
figure [ f : plot_lnb_01 ] shows the extension of our results into the plane , now showing the three - way model comparison .
the left plot still shows the comparison between hz and vary@xmath3 , though neither is the true model except at @xmath41 ( fig . [
f : plot_lnb_01_1d ] is the cross - section of this plot at @xmath42 ) .
the plot is not surprising in the sense that the logarithm of the bayes factor is roughly independent of @xmath6 .
the models hz and vary@xmath3 are just as bad at describing a non - zero running
. however , a slight tilt of the contours appears when @xmath6 goes away from zero .
this indicates that a positive ( resp .
negative ) running can be balanced by a scalar index smaller ( resp .
bigger ) than @xmath43 , accordingly to equation ( [ running ] ) .
this can benefit hz or vary@xmath3 , depending whether it helps or hinders the hz model to fit the data .
in fact the effect just reflects that the scale @xmath44 is not quite at the statistical centre of the data , so that the determination of @xmath1 and @xmath6 has some correlations , and could be removed by judicious choice of the ` pivot ' scale ( corts , liddle & mukherjee 2007 ) .
the centre panel now introduces a comparison of hz with vary@xmath5 , which is the true model in most of the parameter plane . at [ @xmath45=[0,1 ] , the hz model is decisively preferred with @xmath46 .
its higher evidence arises since it can fit the data just as well as vary@xmath5 , but has two less parameters .
once the fiducial point in the two - dimensional space is far enough away from the centre , the hz fit becomes very poor and vary@xmath5 model becomes favoured .
being the true model , vary@xmath5 can simply adapt its two extra free parameters to fit the data at every point of the fiducial space equivalently , thus leading to the same evidence .
we have verified this holds to excellent accuracy in our simulations .
the behaviour of the bayes factor should therefore be approximately symmetrical with respect to @xmath47 and to @xmath42 .
however , it is clearly not quite the case , for the same reason as the presence of the tilt in the left panel .
the influence of the correlation between the two fiducial parameters is greater this time , as it acts on hz only .
the logarithm of the bayes factor , @xmath48 , as a function of the fiducial value of @xmath6 .
the horizontal lines indicate where the comparison becomes ` strong ' ( dashed ) and ` decisive ' ( solid ) on the jeffreys scale.,width=283 ] finally , we need to consider a comparison between the models vary@xmath3 and vary@xmath5 , which is illustrated in the right panel of fig .
[ f : plot_lnb_01 ] .
this plot is fully determined by the above results , as by definition @xmath49 . moreover , for the same reason that the evidence of vary@xmath5 is independent of both fiducial parameters @xmath1 and @xmath6 , vary@xmath3 turns out to be independent of @xmath50 .
this allows us to restrict our analysis to one dimension only , shown in fig .
[ f : plot_lnb_12_1d ] . at @xmath42
the vary@xmath3 model is strongly preferred over vary@xmath5 as @xmath51 , having one less parameter . the running model becomes strongly favoured only if the true running satisfies @xmath52 .
two graphical representations of the three - way model comparison .
the upper panel is a false - colour rgb plot with the probabilities of hz , vary@xmath3 , and vary@xmath5 assigned to the red , green , and blue channels respectively .
the lower panel simply shows the model which would receive the highest model probability at each point in the fiducial parameter space , with those three models allocated white , grey , and black respectively.,title="fig:",width=283 ] + two graphical representations of the three - way model comparison .
the upper panel is a false - colour rgb plot with the probabilities of hz , vary@xmath3 , and vary@xmath5 assigned to the red , green , and blue channels respectively .
the lower panel simply shows the model which would receive the highest model probability at each point in the fiducial parameter space , with those three models allocated white , grey , and black respectively.,title="fig:",width=283 ] in fig .
[ f : rgb ] , we display the full three - way model comparison in two different ways . the three - model case is perfectly adapted to display by false - colour rgb plot , where the intensity of each of the three red green blue colour channels at a given fiducial point is assigned as the posterior model probability , given by bayes theorem , @xmath53 here we assume that the prior model probabilities @xmath54 are equal ( an assumption readily varied if required ) , so the equation simplifies to @xmath55 that the total probability sums to one corresponds to fixed total intensity .
this is shown in the upper panel .
the region which appears red would lead to the hz model being preferred , green the vary@xmath3 model , and blue the vary@xmath5 model . between those , regions which interpolate into secondary colours share their probability between the different models .
there are also four ` vertices ' at which all three models have the same probability .
we see that the transitions between the different domains are rather rapid in terms of the shifting model probabilities .
the lower plot shows a much simpler representation , where regions are shaded simply according to the dominant model probability in that region .
these two plots affirm the results already apparent from the earlier figures ; for planck to be able to demonstrate that @xmath56 , the true value will have to be more than 0.01 away from unity @xcite , and for running to be convincingly detected @xmath57 will need to be at least 0.02 .
according to wmap3 analyses @xcite , the running is presently constrained , at 95% confidence , to be in the range of approximately @xmath58 .
the precise constraints depend on both on the dataset combination used and the model assumptions made ( e.g. whether or not to include tensor perturbations ) , and we have simply quoted the broadest available . although the range is highly skewed to negative values , the special status of @xmath41 , and the prediction from slow - roll inflation for an @xmath6 value that current experiments can not distinguish from zero , means that from a model selection point of view @xmath42 should still be regarded as a very plausible interpretation of the data .
given this inconclusive position , we have addressed the extent to which the planck satellite is likely to resolve the situation , using model selection tools to compare three models : harrison zeldovich ( hz ) , power - law initial perturbations ( vary@xmath3 ) , and the running model ( vary@xmath5 ) .
the expected outcome depends , of course , on which ( if any ) of these models proves to be the correct one .
supposing first that hz is the true model , we found in pahud et al .
( 2006 ) that vary@xmath3 would be strongly , though not decisively , disfavoured after planck .
the present paper adds the new information that the running model would be decisively disfavoured in this circumstance .
suppose instead that vary@xmath3 is true . then vary@xmath3 will be strongly , but not decisively , preferred over vary@xmath5 . however , as shown in pahud et al .
( 2006 ) , the true value of @xmath1 has to be sufficiently far from one in order for vary@xmath3 to be favoured over hz .
depending on the true parameter values , all three models may survive application of planck data .
finally , suppose vary@xmath5 is true .
the alternatives will only be decisively ruled out provided the true value satisfies @xmath59 , otherwise the outcome will again be indecisive .
the conclusion is that planck will improve knowledge as compared to wmap3 , by a factor of around four ( our calculations indicate a projected parameter uncertainty from planck of about @xmath60 on @xmath6 , to be compared with the current @xmath61 from wmap3 alone ) , and thus does have the capability to convincingly detect running if it is prominent .
however , it does not have the accuracy to probe into the region where slow - roll inflation models typically lie ( kosowsky & turner 1995 ) .
our analysis refers to planck satellite data alone , and , as with wmap3 , one would expect some further tightening with incorporation of other datasets probing different length scales . as with any bayesian analysis , the results have some dependence on prior assumptions . for the priors we have chosen on @xmath1 and @xmath6 , the data are able to constrain the likelihood well within them .
consequently , any change in prior ranges that continues to respect this will just change the evidences according to the change in volume , an effect one can readily calculate .
bearing in mind that the jeffreys scale is logarithmic , a sizeable change in prior parameter ranges would be needed to significantly alter the conclusions .
c.p . was supported in part by the swiss sunburst fund , and a.r.l . , p.m. , and d.p . by pparc .
c.p . acknowledges the hospitality of caltech and marc kamionkowski while this work was completed , during a visit supported by the royal astronomical society and by caltech .
acknowledges the hospitality of the institute for astronomy , university of hawaii , while this work was being completed .
we thank jim cline for prompting us to look at this issue .
ashoorioon a. , hovdebo j. l. , mann r. b. , 2005 , nucl .
b , 727 , 63 , gr - qc/0504135 ballesteros g. , casas j. a. , espinosa j. r. , 2006 , jcap , 0603 , 001 , hep - ph/0601134 bastero - gil m. , freese k. , mersini - houghton l. , 2003 , phys . rev .
d , 68 , 123514 , hep - ph/0306289 chen .
c. , feng b. , wang x. , yang z. , 2004 , class .
grav . , 21 , 3223
, astro - ph/0404419 chung d. j. h. , shiu g. , trodden m. , 2003 , phys .
d , 68 , 063501 , astro - ph/0305193 cline j. m. , hoi l. , 2006 , jcap , 0606 , 007 , astro - ph/0603403 corts m. , liddle a. r. , 2006 , phys .
d , 73 , 083523 , astro - ph/0603016 corts m. , liddle a. r. , mukherjee p. , 2007
d , 75 , 083520 , astro - ph/0702170 covi l. , lyth d. h. , melchiorri a. , odman c. j. , 2004 , phys . rev .
d , 70 , 123521 , astro - ph/0408129 easther r. , peiris h. , 2006 , jcap , 0609 , 010 , astro - ph/0604214 gregory p. , 2005
, _ bayesian logical data analysis for the physical sciences _
, cambridge university press jeffreys h. , 1961 , _ theory of probability _ , 3rd ed , oxford university press kawasaki m. , yamaguchi m. , yokoyama j. , 2003 , phys
d , 68 , 023508 , hep - ph/0304161 kosowsky a. , turner m. s. , 1995 , phys .
d , 52 , 1739 , astro - ph/9504071 liddle a. r. , lyth d. h. , 2000 , _
cosmological inflation and large - scale structure _ , cambridge university press liddle a. r. , mukherjee p. , parkinson d. , 2006a , a&g , 47 , 4.30 , astro - ph/0608184 liddle a. r. , mukherjee p. , parkinson d. , wang y. , 2006b , phys . rev .
d , 74 , 123506 , astro - ph/0610126 lidsey j. e. , tavakol r. , 2003 , phys .
b , 575 , 157 , astro - ph/0304113 mackay d. j. c. , 2003 , _ information theory , inference and learning algorithms _ , cambridge university press mukherjee p. , parkinson d. , liddle a. r. , 2006a , apjl , 638 , l51 , astro - ph/0508461 mukherjee p. , parkinson d. , corasaniti p. s. , liddle a. r. , kunz m. , 2006b , mnras , 369 , 1725 , astro - ph/0512484 pahud c. , liddle a. r. , mukherjee p. , parkinson d. , 2006 , phys .
d , 73 , 123524 , astro - ph/0605004 parkinson d. , mukherjee p. , liddle a. r. , 2006 , phys .
d , 73 , 123523 , astro - ph/0605003 skilling j. , 2006 , bayesian anal .
, * 1 * , 833 spergel d. n. et al . ( the wmap team ) , 2003 , apjs , 148 , 175 , astro - ph/0302209 spergel d. n. et al .
( the wmap team ) , 2007 , apjs , 170 , 377 , astro - ph/0603449 trotta r. , 2006 , in proceedings of `` cosmology , galaxy formation and astroparticle physics on the pathway to the ska '' , astro - ph/0607496 trotta r. , 2007a , mnras , 378 , 72 , astro - ph/0504022 trotta r. , 2007b , mnras , 378 , 819 , astro - ph/0703063
|
we use model selection forecasting to assess the ability of the planck satellite to make a positive detection of spectral index running .
we simulate planck data for a range of assumed cosmological parameter values , and carry out a three - way bayesian model comparison of a harrison zeldovich model , a power - law model , and a model including running .
we find that planck will be able to strongly support running only if its true value satisfies @xmath0 .
cosmology : theory
|
this work is concerned with the numerical solution of the time - harmonic maxwell equations discretized by discontinuous galerkin methods on unstructured meshes .
our motivation for using a discontinuous galerkin method is the enhanced flexibility compared to the conforming edge element method @xcite : for instance , dealing with non - conforming meshes is straightforward and the choice of the local approximation space is not constrained .
nonetheless , before taking full advantage of these features , it is required to carefully study the basic ingredients of the method such as the choice of the numerical flux at the interface between neighboring elements . in the context of time - harmonic problems ,
the design of efficient solution strategies for the resulting sparse linear systems is an equally important question .
previous works have shown convergence results for discontinuous galerkin methods applied to the time - harmonic maxwell equations , studied in the form of second - order vector wave equations .
most of these works use a mixed formulation @xcite but discontinuous galerkin methods on the non - mixed formulation have recently been proved to converge ( interior penalty technique @xcite as well as the local discontinuous galerkin method @xcite ) .
the convergence properties of these methods in the time - domain case have been studied in @xcite when using a centered flux and in @xcite when using an upwind flux .
the case of the upwind flux has been analyzed in @xcite and @xcite for the time - harmonic problems and the convergence has been proved only for a perturbed problem . the general case of friedrichs systems and the elliptic maxwell equations in particular has been treated in @xcite and @xcite . however , to our knowledge , no direct convergence analysis on the first - order time - harmonic system has been conducted so far , which should be useful , for instance , when using an upwind flux ( see subsection [ sec : flux_num ] ) .
the main contribution of this work is a numerical study of the convergence of discontinuous galerkin methods based on centered and upwind fluxes applied to the first - order time - harmonic maxwell system in the two - dimensional case .
the paper is organized as follows : in section [ sec : form ] we present the discretization method as well as the different kind of fluxes considered . in section [ sec : convdis ] , the convergence properties are recalled in the case of the elliptic maxwell equations and the solvability of the discrete perturbed problem is analyzed in the case of the centered flux . in section [ sec : res_2d ] the numerical convergence is studied and confronted to the theoretical convergence order .
different numerical fluxes are then compared on two distinct examples .
the system of non - dimensionalized time - harmonic maxwell s equations can be written in the following form : @xmath0 where @xmath1 and @xmath2 are the unknown electric and magnetic fields and @xmath3 is a known current source . the parameters @xmath4 and @xmath5 are respectively the complex - valued relative dielectric permittivity ( integrating the electric conductivity ) and the relative magnetic permeability ; we consider here the case of linear isotropic media .
the angular frequency of the problem is given by @xmath6 .
we solve equations in a bounded domain @xmath7 , and on its boundary @xmath8 , we impose the following boundary conditions : @xmath9 the vectors @xmath10 and @xmath11 represent the components of an incident electromagnetic wave .
we can further rewrite + , assuming @xmath3 equals to 0 , under the following form : @xmath12 where @xmath13 is the new unknown vector and @xmath14 denoting by @xmath15 the canonical basis of @xmath16 , the matrices @xmath17 with @xmath18 are given by : @xmath19 in the following we denote by @xmath20 the sum @xmath21 and by @xmath22 and @xmath23 its positive and negative parts is the eigenfactorization then @xmath24 where @xmath25 ( resp .
@xmath26 ) only gathers the positive ( resp .
negative ) eigenvalues . ] .
we also define @xmath27 . in order to take into account the boundary conditions , the matrices @xmath28 and @xmath29
are given by : @xmath30 see @xcite for further details on the derivation of this formulation .
let @xmath31 denote a discretization of the domain @xmath7 into a union of conforming elements ( tetrahedral or hexahedral elements ) @xmath32 we look for the approximate solutions @xmath33 of in @xmath34 where the function space @xmath35 is defined by : @xmath36 ^ 3 \ / \ \forall k \in { \mathcal{t}}_h , \ \ { \mbox{\boldmath $ v$}}_{\lvert k } \in { \mathcal{p}}(k ) \right\}.\ ] ] the term @xmath37 denotes a space of polynomial functions on the element @xmath38 .
we take the scalar product of the first equation of ( [ eq:3dformfort ] ) by a sufficiently smooth vector field @xmath39 and we integrate over an element @xmath38 of the mesh @xmath40 : @xmath41 by using green s formula we obtain a weak formulation involving a boundary term .
this term is replaced in discontinuous galerkin methods by a function @xmath42 which is usually referred as the numerical flux ( see also ern and guermond @xcite ) ; the aim is then to determine @xmath43 in @xmath34 such that : @xmath44 in order to couple the element @xmath38 with its neighbors for ensuring the consistency of the discretization , this numerical flux can be defined in the following way : @xmath45 where @xmath46 , @xmath47 and @xmath48 respectively denote the set of interior faces , the set of faces on @xmath49 and the set of faces on @xmath50 .
@xmath51 stands for the incidence matrix between oriented faces and elements whose entries are given by : @xmath52 we also define respectively the jump and the average of @xmath39 on a face @xmath53 shared by two elements @xmath38 and @xmath54 : @xmath55 finally , the matrix @xmath56 allows to penalize the jump of a field or of some components of this given field on the face @xmath53 and the matrix @xmath57 to be defined later insures the asymptotic consistency with the boundary conditions of the continuous problem . in this study
, we aim at comparing the properties of three classical numerical fluxes : - * a centered flux * ( see @xcite for the time - domain equivalent ) . in this case
@xmath58 for all the faces @xmath53 and , for the boundary faces , we use : @xmath59 - * an upwind flux * ( see @xcite ) . in this case : @xmath60 with @xmath61 , @xmath62 and @xmath63 equals to @xmath64 for homogeneous media .
the definition of @xmath65 for @xmath53 in @xmath47 is identical to the centered case .
- * a partially penalized upwind flux * ( local discontinuous galerkin method , see @xcite ) .
this flux is characterized by a penalization coefficient given by : @xmath66 the definition of @xmath65 for @xmath53 in @xmath47 is also identical to the centered case .
we are interested in assessing these numerical fluxes for the discretization of .
firstly , we want the best asymptotic convergence order in @xmath67-norm for the electric and magnetic field for a fixed polynomial order approximation on an unstructured mesh . secondly , a minimal numerical dispersion is also needed . in the following
we will focus on the first criterion .
the asymptotic convergence order in @xmath67-norm between the exact solution @xmath68 and the approximate solution @xmath69 corresponds to the largest real coefficients @xmath70 and @xmath71 such that : @xmath72 where @xmath73 is the mesh size .
let us note that in the numerical examples proposed in section [ sec : res_2d ] , we will often equivalently consider the evolution of the norm of the error against the square root of the number of degrees of freedom ( dofs ) , in order to deduce coefficients @xmath70 and @xmath71 .
we first recall in table [ tab : max_ell ] below the theoretical convergence order for the elliptic maxwell equations @xcite , for a sufficiently smooth solution and when the local function space @xmath37 is @xmath74 ^ 3 $ ] _ i.e. _ the space of vectors whose components are polynomials of order at most @xmath75 . when using the flux with a penalization of @xmath1 , similar convergence results are proved for the time - harmonic maxwell equations in @xcite . on figure [ fig : compare_tu ]
, we compare the evolution of the @xmath67-norm of the error with the mesh size @xmath73 by using the meshes @xmath76 , for both a centered flux and an upwind flux , figure [ fig : competu ] corresponds to the error for the field @xmath1 while figure [ fig : comphtu ] corresponds to the error for the field @xmath2 .
the results for the upwind flux are the same as for the uniformly refined meshes .
for the centered flux , note the lack of convergence for the case @xmath77 . for all the other cases
the results remain the same as for the uniformly refined meshes .
it is already known for time - domain problems that the centered flux combined to a leap - frog time integration scheme results in a non - dissipative discontinuous galerkin method ( a mandatory feature for long time computations , see @xcite ) . as far as time - harmonic problems are concerned , the previous results show that the upwind flux has better convergence properties .
nevertheless , the centered flux remains less expensive both for time - domain and time - harmonic problems ( arithmetic operations and memory requirements ) .
the domain is the square @xmath78 ^ 2 $ ] where we have suppressed a part by inserting a point of coordinates @xmath79 at it is shown on figure [ fig : num2_meshes ] .
the properties @xmath80 and @xmath5 are still homogeneous and equal to one.appropriate non - homogeneous dirichlet boundary conditions are enforced on the boundary of the domain in order to obtain @xmath81 as the solution .
loula fezoui , stphane lanteri , stphanie lohrengel , and serge piperno . convergence and stability of a discontinuous galerkin time - domain method for the 3d heterogeneous maxwell equations on unstructured meshes .
, 39(6):11491176 , 2005 .
|
we present numerical results concerning the solution of the time - harmonic maxwell equations discretized by discontinuous galerkin methods . in particular ,
a numerical study of the convergence , which compares different strategies proposed in the literature for the elliptic maxwell equations , is performed in the two - dimensional case . ,
, , time - harmonic maxwell s equations , discontinuous galerkin methods .
|
a graph , @xmath3 , is _ intrinsically knotted _ if every embedding of @xmath3 in @xmath4 contains a nontrivial knot .
a link @xmath5 is _ splittable _ if there is an embedding of a 2-sphere @xmath6 in @xmath7 such that each component of @xmath8 contains at least one component of @xmath5 .
if @xmath5 is not splittable it is called _ non - split_. a graph , @xmath3 , is _ intrinsically linked _
if every embedding of @xmath3 in @xmath4 contains a non - split link . a graph , @xmath3 , is _ minor minimal with respect to being intrinsically linked _ ( or simply _ minor minimal intrinsically linked _ ) if @xmath3 is intrinsically linked and no minor of @xmath3 is intrinsically linked .
the combined work of conway and gordon @xcite , sachs @xcite , and robertson , seymour , and thomas @xcite fully characterizes intrinsically linked graphs .
the graphs in the petersen family , shown in figure [ pf ] , are the complete set of minor minimal intrinsically linked graphs .
so no minor of one of the graphs of the petersen family is intrinsically linked , and every graph that is intrinsically linked contains one of these graphs as a minor .
let the set of the seven graphs of the petersen family be denoted by @xmath9 .
the concept of a graph being intrinsically linked can be generalized to a graph that intrinsically contains a link of more than two components .
a graph @xmath3 is _ intrinsically @xmath10-linked _ if every embedding of @xmath3 in @xmath4 contains a non - split @xmath10-component link . from here forward
we will use _
@xmath10-link _ to mean a non - split @xmath10-component link . in this paper
we focus on intrinsically 3-linked graphs . in section [ back ] , we discuss the set of known intrinsically 3-linked graphs and introduce our five new constructions . in section [ i3-l ] , we prove that each of the new constructions results in an intrinsically 3-linked graph .
( 260 , 330 ) ( 0,0 ) moves .
, title="fig : " ] ( 65,305)@xmath11 ( 190,310)@xmath12 ( 125,250)@xmath13 ( 245,220)@xmath14 ( 180,100)@xmath15 ( 65,130)@xmath16 ( 130,15)@xmath17 ( 200 , 70 ) ( 0,-10 ) move and @xmath18 move.,title="fig : " ] ( 72 , 41)*@xmath0 move * ( 72 , 20)*@xmath19 move * a _ @xmath20 move _ on an abstract graph is where a valance 3 vertex , @xmath21 , together with its adjacent edges are deleted , and three edges are added , one between each pair of vertices that had been adjacent to @xmath21 .
the reverse move is called a _
@xmath18 move_. see figure [ ytri ] . in @xcite ,
sachs showed that each graph in the petersen family , i.e. all those graphs obtained from @xmath11 by @xmath0 and @xmath19 moves , is also minor minimal intrinsically linked .
motwani , raghunathan , and saran @xcite showed that both intrinsic linkedness and intrinsic knottedness are preserved by @xmath22 moves .
their proof that intrinsic linkedness is preserved by @xmath22 moves immediately generalizes to show that intrinsic @xmath10-linkedness is also preserved by @xmath22 moves .
robertson , seymour , and thomas @xcite showed that @xmath0 moves also preserve intrinsic linkedness . on the other hand , flapan and
naimi @xcite showed that @xmath0 moves do not preserve intrinsic knottedness .
it is not known if intrinsic @xmath10-linkedness is preserved by @xmath0 moves in general .
the work in @xcite showed that intrinsic @xmath23-linkedness is preserved by @xmath0 moves .
while the family of minor minimal intrinsically linked graphs ( also minor minimal intrinsically 2-linked graphs ) is connected by @xmath0 and @xmath19 moves , the family of minor minimal intrinsically 3-linked graphs is not @xcite .
it is also known that if the graph resulting from a @xmath0 move on a minor minimal intrinsically @xmath10-linked graph is intrinsically @xmath10-linked , then it is minor minimal intrinsically @xmath10-linked @xcite . in section [ ytri ] , we prove that intrinsic 3-linkedness is not preserved by @xmath0 moves .
the author would like to thank dorothy buck , erica flapan , kouki taniyama and r. sean bowman for helpful conversations and their continued support .
there are a number of graphs already known to be intrinsically 3-linked .
figure [ i3l ] shows all those graphs that have been shown to be intrinsically 3-linked , where no minor is known to be intrinsically 3-linked ( only @xmath24 is known to be minor minimal intrinsically 3-linked ) . in @xcite , flapan , naimi , and pommersheim investigate intrinsically 3-linked graphs ( or _ intrinsically triple linked _
they proved that the complete graph on ten vertices , @xmath25 is the smallest complete graph to be intrinsically 3-linked .
bowlin and foisy @xcite also looked at intrinsically 3-linked graphs .
they exhibited two different subgraphs of @xmath25 that are also intrinsically 3-linked , the graph resulting from removing two disjoint edges from @xmath25 , call it @xmath26 , and the graph obtained by removing four edges incident to a common vertex from @xmath25 , call it @xmath27 .
so @xmath25 is not minor minimal intrinsically 3-linked .
they also described two constructions that give intrinsically 3-linked graphs , the first being the graph that results from identifying an edge of @xmath28 with an edge of @xmath29 , when @xmath28 and @xmath29 are either @xmath30 or @xmath31 ; we will call this graph @xmath32 . since all of the edges of @xmath30 are equivalent as are those of @xmath31 this gives rise to three graphs .
one of these graphs , @xmath33 ( also called @xmath34 ) , was previously shown to be intrinsically 3-linked in @xcite .
the second is the graph obtained by connecting two graphs @xmath35 by a 6-cycle where the vertices of the 6-cycle alternate between @xmath28 and @xmath29 .
we will call such a graph @xmath36 . the subscript @xmath37 is given because there can be multiple ways to combine the same two graphs in this way .
not all of the vertices of each of the graphs of the petersen family are equivalent , so there are many different ways to form the 6-cycle .
thus , there will be many more than @xmath38 such graphs that one might first expect from combining the seven different graphs of @xmath9 in this construction . due to the large number of graphs of this form they are not drawn in figure [ i3l ] but instead
a pictorial representation of any such graph is shown .
flapan , foisy , naimi , and pommersheim addressed the question of minor minimal intrinsically @xmath10-linked graphs in @xcite , where they constructed a family of minor minimal intrinsically @xmath10-linked graphs .
the minor minimal intrinsically @xmath39-linked graph they constructed was called @xmath24 , shown in figure [ i3l ] .
( 260 , 330 ) ( 0,0 ) is known to be minor minimal.),title="fig : " ] ( 33,215)@xmath26 ( 180,207)@xmath27 ( 50,144)@xmath24 ( 175,148)@xmath40 ( 25,7)*@xmath41 * ( 109,11)*@xmath42 * ( 178,23)*@xmath43 * in section [ i3-l ] , we prove that the following five constructions give rise to intrinsically 3-linked graphs : let @xmath44 and let @xmath45 be a vertex in the graph @xmath46 .
let the set of adjacent vertices to the vertex @xmath45 be @xmath47 , for @xmath48 .
: let the graph obtained by adding the edges between @xmath49 and all but one of the vertices of @xmath50 and the edges between @xmath51 and all but one of the vertices of @xmath52 to the graphs @xmath28 and @xmath29 be called @xmath53 .
we call this the _ double star construction_. : let the graph obtained by identifying the vertices @xmath49 and @xmath51 , adding a vertex @xmath54 and the edges from @xmath54 to all but one of the vertices in the set @xmath52 and all but one of the vertices in the set @xmath50 be called @xmath55 . :
the graph @xmath16 has two vertices of valence three , label them @xmath54 and @xmath56 .
let @xmath3 be one of the graphs of the petersen family , let @xmath21 be one of the vertices of @xmath3 , and let @xmath57 be the set of vertices adjacent to @xmath21 in @xmath3 . let the graph obtained by identifying the two vertices @xmath54 and @xmath21 , and adding edges between @xmath56 and all but one of the vertices in @xmath57 be called @xmath58 .
: let the _ vertex identification construction _ be the construction where a graph @xmath59 is formed by adding edges between the sets of vertices @xmath52 and @xmath50 , such that between every pair of vertices of @xmath52 and pair of vertices of @xmath50 there is at least one edge joining a vertex from @xmath52 to a vertex from @xmath50 and then identifying the vertices @xmath49 and @xmath51 to get a single vertex @xmath54 .
let the set of added edges between @xmath52 and @xmath50 be @xmath60 , where @xmath61 and @xmath62 .
: let @xmath63 be the full vertex set of @xmath46 .
let @xmath64 be the graph obtained by adding five disjoint edges between @xmath65 and @xmath66 to the graphs @xmath28 and @xmath29 .
the vertex is dropped from the notation for constructions 1 , 2 and 3 , if the vertices of the graph @xmath28 or @xmath29 are equivalent .
these constructions introduce numerous new intrinsically 3-linked graphs .
for example , the graphs @xmath67 , @xmath68 are both subgraphs of @xmath41 .
see figure [ mm ] .
so these two graphs together with all of the graphs that can be obtained from them by @xmath22 moves were previously unknown to be intrinsically 3-linked .
similarly , @xmath69 is a subgraph of @xmath42 .
so the graph @xmath69 introduces a set of new intrinsically 3-linked graphs .
in this section we prove that the five new constructions explained in section [ back ] give intrinsically 3-linked graphs .
these constructions exploit some nice properties of the graphs in the petersen family .
for each @xmath70 every pair of disjoint cycles contains all of the vertices of @xmath3 .
thus every embedding of a graph from the petersen family in @xmath4 not only contains a 2-link but contains a 2-link which contains all of the vertices of the graph .
it is known that , for any @xmath70 , every embedding of @xmath3 in @xmath4 contains a two component link with odd linking number @xcite .
so we will work with linking mod ( 2 ) and denote the mod ( 2 ) linking number of two simple closed curves @xmath5 and @xmath34 by @xmath71 .
, ( b ) @xmath68 , and ( c ) @xmath69 . ]
we will use the following lemma , proved in @xcite , to prove that construction 1 of the previous section gives rise to an intrinsically 3-linked graph .
[ 2path ] in an embedded graph with mutually disjoint simple closed curves , @xmath72 , @xmath73 , @xmath74 and @xmath75 , and two disjoint paths @xmath76 and @xmath77 , such that @xmath76 and @xmath77 begin in @xmath73 and end in @xmath78 , if @xmath79 then the embedded graph contains a non - splittable 3-component link .
[ starstar ] let @xmath53 be a graph obtained via construction 1
. then @xmath53 is intrinsically 3-linked .
fix an arbitrary embedding of @xmath53 .
since @xmath28 is a graph in the petersen family we know it must contain a 2-link and that @xmath49 must be in one of the components of the 2-link .
let the component that contains @xmath49 be called @xmath73 and the other component be called @xmath72 .
note that @xmath80 .
similarly , @xmath29 must contain a 2-link and that @xmath51 must be in one of the components of the 2-link .
let the component that contains @xmath51 be called @xmath78 and the other component be called @xmath75 , note that @xmath81 . since @xmath49 is in @xmath73 two of the vertices adjacent to @xmath49 are also in @xmath73 .
at least one of these vertices must be adjacent to @xmath51 , call it @xmath82 .
note , the edge @xmath83 goes between @xmath73 and @xmath78 .
next , since @xmath51 is in @xmath78 two of the vertices adjacent to @xmath51 are also in @xmath78 and at least one of them is adjacent to @xmath49 , call it @xmath84 .
the edge @xmath85 also goes between @xmath73 and @xmath78 .
notice that @xmath83 and @xmath85 can not be the same edges by construction .
thus by lemma [ 2path ] we see that the chosen embedding contains a 3-link , and so @xmath53 is intrinsically 3-linked .
to prove proposition [ con2 ] , we will use the following lemma which appears in @xcite : [ 3lpath ] suppose that @xmath3 is a graph embedded in @xmath4 and contains the simple closed curves @xmath86 and @xmath75 .
suppose that @xmath72 and @xmath75 are disjoint from each other and both are disjoint from @xmath73 and @xmath78 , and that @xmath73 and @xmath78 intersect in precisely one vertex @xmath54 .
also , suppose there are vertices @xmath87 in @xmath73 and @xmath88 in @xmath78 and a path p in @xmath3 with endpoints @xmath89 and @xmath21 whose interior is disjoint from each @xmath90 .
if @xmath79 , then there is a non - splittable 3-component link in @xmath3 .
[ con2 ] let @xmath55 be a graph obtained via construction 2 , then @xmath55 is intrinsically 3-linked .
let @xmath47 be the sets vertices and @xmath54 be the vertex as described in construction 2 , in the previous section .
fix an arbitrary embedding of @xmath55 .
since @xmath28 is in the petersen family we know it must contain a 2-link and that @xmath49 must be in one of the components of the 2-link .
let the component that contains @xmath49 be called @xmath73 and the other component be called @xmath72 ; note that @xmath80 .
similarly , @xmath29 must contain a 2-link and that @xmath51 must be in one of the components of the 2-link .
let the component that contains @xmath51 be called @xmath78 and the other component be called @xmath75 ; note that @xmath81 . since @xmath49 is in @xmath73 there are two vertices in @xmath52 that are also in @xmath73 and at least one of them is adjacent to @xmath54 .
call it @xmath91 . similarly , since @xmath51 is in @xmath78 there are two vertices of @xmath50 that are also in @xmath78 and at least one of them is adjacent to @xmath54 .
call it @xmath92 .
let the path consisting of the two edges @xmath93 and @xmath94 be called @xmath95 .
the path @xmath95 goes from @xmath73 to @xmath78 .
so by lemma [ 3lpath ] we see that the embedding contains a 3-link .
thus @xmath55 is intrinsically 3-linked .
the graph @xmath69 is shown in figure [ mm ] .
we will use the following lemmas in the proof of the next proposition about construction 3 . see section [ back ] for the constructions .
[ k44 ] @xcite let @xmath31 be embedded in @xmath4 , then every edge of @xmath31 is in a component of a 2-link .
[ 3l ] @xcite suppose that @xmath3 is a graph embedded in @xmath4 that contains the simple closed curves @xmath86 and @xmath75 .
suppose that @xmath72 and @xmath75 are disjoint from each other and both are disjoint from @xmath73 and @xmath78 , and that @xmath96 is an arc .
if @xmath80 and @xmath81 , then there is a non - split 3-component link in @xmath3 .
let @xmath58 be a graph obtained via construction 3 , then @xmath58 is intrinsically 3-linked .
let @xmath54 , @xmath56 , and @xmath57 be as described in construction 3 .
fix an arbitrary embedding of @xmath58 .
since @xmath3 is one of the graphs from the petersen family it contains a 2-link @xmath97 which contains the vertex @xmath98 , without loss of generality let @xmath54 be in @xmath73 . since @xmath54 is in @xmath73 two of the vertices adjacent to @xmath54 , are also in @xmath73 .
so at least one of these vertices in @xmath73 is adjacent to @xmath56 , call the vertex @xmath82 . label the path p , that is comprised of the two edges @xmath99 and @xmath100 .
notice @xmath101 and @xmath102 .
now the subgraph @xmath16 together with the path @xmath95 form a subdivision of @xmath31 , i.e. this can be viewed as @xmath31 where @xmath95 is one if the edges . by lemma [ k44 ] for every embedding of @xmath31 each edge
is contained in a 2-link , so @xmath95 is contained in a 2-link @xmath103 . without loss of generality ,
let @xmath95 be an edge of @xmath78 .
so @xmath104 is an arc , @xmath80 and @xmath81 .
thus by lemma [ 3l ] the embedding of @xmath58 contains a 3-link . and @xmath62 .
this figure shows possible sets of added edges @xmath60 for the vertex identification constructions for all different possible sizes of the vertex set @xmath52 and @xmath50 .
for each of them the vertex sets @xmath52 and @xmath50 are shown vertically and additional edges @xmath60 are shown . ]
recall construction 4 , the vertex identification construction , where a set of edges @xmath60 is added between the two sets of vertices @xmath52 and @xmath50 . for the full definition refer to section [ back ] .
the graphs in the petersen family have vertices of valence 3 , 4 , 5 , and 6 .
let @xmath61 and @xmath62 .
we want to construct @xmath60 , a set of edges between @xmath52 and @xmath50 such that given any pair of vertices from @xmath52 and any pair of vertices from @xmath50 there is an edge between two of the vertices from the chosen pair .
so each pair of vertices from @xmath52 must be connected to @xmath105 vertices from @xmath50 . to reduce the total number of edges needed we divide the edges evenly between the two vertices , so each vertex of @xmath52 is connected to @xmath106 vertices of @xmath50 .
suppose @xmath107 , this gives a lower bound of @xmath108 , if @xmath109 is odd , and @xmath110 if @xmath109 is even .
figure [ addedge ] shows possible sets of edges @xmath60 to be added between @xmath52 and @xmath50 for all possible combinations of valence . it can be checked that there sets of vertices satisfy the criterion . however these are not the only possible @xmath60 sets , and it is not known if they are optimal . in the case of @xmath111
then @xmath112 is the lower bound but in all other examples given @xmath113 is greater than the lower bound obtained .
[ vidi3l]any graph @xmath59 constructed through the vertex identification construction of graphs @xmath28 and @xmath29 in the petersen family is intrinsically 3-linked .
consider an arbitrary embedding of @xmath59 .
let the notation be as in construction 4 : the identified vertex is labelled @xmath54 , the sets of vertices adjacent to @xmath54 in the subgraphs @xmath28 and @xmath29 , respectively , are labelled @xmath52 and @xmath50 , and the set of edges between @xmath52 and @xmath50 is @xmath60 . since every vertex of a petersen graph is contained in every link in the embedding
, @xmath54 is in one of the components of the link in each @xmath46 .
let the components of the link in @xmath28 be labelled @xmath72 and @xmath73 , with the vertex @xmath54 in the component @xmath73 , and let the components of the link in @xmath29 be labelled @xmath78 and @xmath75 , with the vertex @xmath54 in the component @xmath78 .
a pair of vertices from the set @xmath65 is also part of @xmath73 and , similarly , a pair of vertices from the set @xmath66 is also part of @xmath78 . by construction , there is an edge , @xmath114 , of the set @xmath60 between @xmath73 and @xmath78 .
since none of the edges of @xmath60 are contained in @xmath28 or @xmath29 , the interior of the edge @xmath114 is disjoint from the links @xmath97 and @xmath103 .
thus , by lemma [ 3lpath ] , h contains a 3-link .
the graph @xmath64 obtained by construction 5 , is intrinsically 3-linked .
let @xmath63 be the vertex set of @xmath46 , let the set of five added edges be @xmath115 .
fix an embedding of @xmath64 .
since @xmath35 , @xmath28 contains a 2-link that contains all of the vertices of @xmath65 , call the link @xmath97 , and @xmath29 contains a 2-link that contains all of the vertices of @xmath66 , call the link @xmath103 . because all of the vertices are in one of the 2-links each edge of @xmath115 will go between components of the different 2-links .
there are four different pairs of components that can be connected by the said edges , so by the pigeonhole principle two of the edges must go between the same pair of components . by lemma [ 2path ]
the embedding contains a 3-link .
thus @xmath64 is intrinsically 3-linked .
in this section , we show that intrinsic 3-linkedness is not preserved by @xmath0 moves .
( 260 , 130 ) ( 0,10 ) .
an embedding of @xmath1 that does not contain a 3-link .
, title="fig : " ] ( 25,88)*1 * ( 45,94)*2 * ( 63,104)*3 * ( 107,90)*4 * ( 107,72)*5 * ( 107,50)*6 * ( 65,7)*7 * ( 45,20)*8 * ( 25,35)*9 * ( 150,108)*@xmath82 * ( 200,118)*@xmath84 * ( 133,72)*@xmath116 * ( 180,93)*@xmath117 * ( 230,72)*@xmath114 * ( 157,55)*@xmath118 * ( 187,75)*@xmath119 * ( 203,57)*@xmath120 * ( 150,23)*@xmath121 * ( 207,23)*@xmath122 * we begin with @xmath2 , since all of the vertices of @xmath123 are equivalent there is a single graph that can be obtained throughout the double star construction with two petersen graphs . by proposition [ starstar ] , @xmath2 is intrinsically 3-linked .
let the graph obtained by a @xmath0 move on @xmath2 as indicated in figure [ ytri ] be called @xmath1 .
we claim that , the graph @xmath1 is not intrinsically 3-linked . consider the embedding @xmath124 shown in figure [ f(m ) ] .
let the vertices be labelled as indicated .
let @xmath125 be the embedded subgraph defined by the vertices 1 , 2 , 3 , 4 , 5 , 6 , 7 , 8 , 9 and the edges between them in @xmath124 , and let @xmath126 be the embedded subgraph defined by the vertices @xmath127 and the edges between them .
for @xmath124 to contain a 3-link , the 3-link must be in both the embedded subgraphs @xmath125 and @xmath126 , since neither contains three disjoint simple closed curves on their own .
since @xmath125 and @xmath126 are disjoint and there is no linking between them , two of the edges joining the subgraphs must also be in the 3-link .
the subgraph @xmath125 contains a single linked pair of cycles , indicated with thickened edges .
similarly , in @xmath126 there is a single linked pair of cycles , indicated with thickened edges .
all other cycles in the subgraphs @xmath125 and @xmath126 bound disks that do not intersect the graph in their interiors .
no pair of edges between the two subgraphs @xmath125 and @xmath126 connects two of the linked cycles .
so there is no 3-link in @xmath124 .
notice that there are many graphs that can be constructed with the double star construction that are a @xmath0 move away from a graph that is not intrinsically 3-linked .
consider @xmath128 for any @xmath70 , a similar @xmath0 move to that is the proof above will produce a graph that is not intrinsically 3-linked .
more generally , this can be done with any double star construction where the vertex of @xmath47 that is not connected to the vertex @xmath129 is trivalent .
h. sachs , _ on a spatial analogue of kuratowski s theorem on planar graphs an open problem _ , graph theory , lag@xmath131w , 1981 , lecture notes in mathematics ,
1018 ( springer - verlag , berlin , heidelberg , 1983 ) , 649662 .
|
this paper introduces a number of new intrinsically 3-linked graphs through five new constructions .
we then prove that intrinsic 3-linkedness is not preserved by @xmath0 moves .
we will see that the graph @xmath1 , which is obtained through a @xmath0 move on @xmath2 , is not intrinsically 3-linked .
|
at small layer separations , the ground state of a @xmath0 bilayer quantum hall system exhibits spontaneous interlayer phase coherence .
interest in this issue has been renewed by intriguing tunneling transport measurements by spielman _
_ showing a very pronounced conductance peak at zero bias voltage @xcite for sufficiently small values of the ratio of layer separation to magnetic length .
the quantum phase transiton underlying this phenomenon has been investigated recently in an exact diagonalization study using the spherical geometry @xcite .
the results suggest that a single phase transition , likely of first order , separates incompressible states with strong interlayer correlations from compressible states with weak interlayer correlations . in the present work we report on a study of @xmath0 quantum hall bilayer using a different approach introducing an effective spin - pseudospin model on an imaginary lattice in the lowest landau level ( lll ) orbital space
( von neumann lattice ) .
we also add further exact diagonalization results concerning the position and order of the compressible - incompressible transition .
a single - particle state of electron in a @xmath0 bilayer is specified by three quantum numbers : lowest landau level ( lll ) orbit - center quantum number @xmath1 , spin @xmath2 and pseudospin @xmath3 , describing the layer degree of freedom . however , in the incompressible state bilayer has a gap for charged excitations and therefore one can assume that only the spin and pseudospin degrees of freedom are relevant for the low temperature physics .
the microscopic hamiltonian reads @xmath4 , where @xmath5 represents the usual coulomb interaction within ( @xmath6 ) and between layers ( @xmath7 ) , and the single - particle hamiltonian @xmath8 implements tunneling of electrons between the layers with an amplitude @xmath9 and couples electron spins to the perpendicular magnetic field with amplitude @xmath10 .
one would like to eliminate the irrelevant charge degree of freedom from the microscopic hamiltonian and arrive at an effective model containing only spin and pseudospin variables . the most convenient and mathematically rigorous way to do this
is provided by the functional integral approach .
one notices that the no - charge - fluctuations subspace of the system is spanned by single slater determinant many body wave functions of the form @xmath11\rangle = \prod_i \left(\sum_{k=1}^4 z_{ik } c^{\dagger}_{ik } \right)| 0 \rangle$ ] . here
@xmath1 is the lll orbital quantum number and @xmath12 is a 4-component spinor index describing the combined spin - pseudospin degree of freedom .
@xmath13 means electron in the top layer with an up spin , @xmath14top layer down spin , @xmath15bottom layer up spin , @xmath16bottom layer down spin .
@xmath17 is the creation operator for an electron in the lll orbital state @xmath1 and with 4-spinor index @xmath12 .
@xmath18 are complex amplitudes , satisfying the normalization condition @xmath19 one can therefore formally write the partition function of the system as a functional integral over this overcomplete set of slater determinants . after considerable algebra one arrives at the following effective hamiltonian @xmath20 \nonumber\\ & + & \sum_{ij } \big [ ( 2h_{ij}-\frac{1}{2}f^s_{ij})t^z_it^z_j- \frac{1}{2}f^d_{ij}{\bf t}^{\perp}_i { \bf t}^{\perp}_j\nonumber\\ & -&\frac{1}{2}f^s_{ij}{\bf s}_i { \bf s}_j 2 f^s_{ij}({\bf s}_i { \bf s}_j ) t^z_i t^z_j\nonumber\\ & -&2f^d_{ij}({\bf s}_i { \bf s}_j)({\bf t}^{\perp}_i { \bf t}^{\perp}_j)\big]\end{aligned}\ ] ] here @xmath21 , and @xmath22 are the direct and exchange matrix elements of the coulomb interaction @xmath23 , and @xmath24 and @xmath25 are local spin and pseudospin operators , respectively .
our spin - pseudospin model is defined on a `` lattice '' with `` sites '' labeled by the lll orbital quantum numbers @xmath1 .
so far we have not specified the orbital basis we are using .
it is obvious that the usually used orbit - center quantum numbers in landau or symmetric gauges are not good choices here .
both of these basis sets introduce an artificial gauge - dependent asymmetry into the problem .
one would like to use a basis more appropriate for a spin model : that of wannier - like functions localized at the sites of imaginary square lattice ( von neumann lattice ) with lattice constant @xmath26 to accomodate exactly one electron or corresponding spin - pseudospin pair per site at filling factor @xmath0
. it is not obvious in advance that such a basis exists .
strong magnetic field imposes certain restrictions on the localization properties of magnetic orbitals @xcite , and it is well established for example that a set of linearly independent _ and _ exponentially localized single - particle orbitals in the lll does not exist
. however , it turns out to be possible @xcite to construct a complete orthonormal set of wannier - like eigenfunctions , which , although not exponentially localized , have a well defined gaussian core with power law falloff at large distances . following @xcite we will call them magnetic wannier functions .
the procedure one uses to construct such a basis set is very much like the one used to construct the usual wannier functions in a crystal .
one starts from the set of minimum uncertainty wavepackets for electrons in the lll , centered at the sites of the square lattice described above .
the difference from the case of a crystal here is that this set is _ overcomplete _ , as was shown by perelomov @xcite .
one then constructs bloch functions from linear combinations of the minimum uncertainty wavepackets and fourier transforms them to obtain the wannier functions .
there are subtleties in this procedure and we refer the reader to the original papers @xcite for further details .
in this paper we will present only the simplest linear spin wave results for our spin - pseudospin model , which are applicable at zero temperature .
a detailed study of finite temperature properties of ( [ 6 ] ) will be presented in a forthcoming publication @xcite . as evident by direct inspection ,
the interaction part of ( 6 ) has the correct @xmath27 symmetry .
correspondingly there are two goldstone modes associated with the spontaneous breaking of this symmetry .
their dispersions are given by @xmath28 in fig.1 we show dispersions ( [ 7 ] ) evaluated for a 20@xmath2920 square lattice using realistic values for the tunneling and zeeman splittings at interlayer separation @xmath30 with @xmath31 being the magnetic length .
one can see that the spin dispersion is quadratic and the pseudospin one is linear at small values of the wavevector in accordance with broken @xmath32 symmetry .
another thing to notice is that the pseudospin gap is appreciably larger than the spin gap even though the bare values of tunneling and zeeman coupling are the same .
this makes invalid ( at finite temperature ) the usual argument about spin fluctuations being frozen out by the magnetic field @xcite .
the dip in the pseudospin mode dispersion at the brillouin zone boundary signals softening of the pseudospin mode due to the development of antiferromagnetic instability , which at large enough interlayer separations ( @xmath33 in our model ) leads to the transition to the compressible state .
our exact diagonalization results are obtained for a bilayer quantum hall system in the spherical geometry containing spin - polarized electrons .
we will consider both the case of zero and finite width @xmath34 of the two quantum wells @xcite . in the exact diagonalization data a transition between a compressible ground state with weak interlayer correlations to an incompressible strongly correlated phase is signalled by a maximum of the fluctuation @xmath35 of the ground state pseudospin magnetisation along with its susceptibility @xmath36 @xcite .
these maxima grow rapidly with increasing system indicating a quantum phase transition .
the positions of these maxima as a function of tunneling gap and layer separation ( measured in units of the magnetic length ) define finite - size phase boundaries ( cf . figure 3 in ref .
the spectacular phenomenon found by spielman
_ et al . _
@xcite occured in samples with extremely small tunneling amplitude close to the limit of vanishing tunneling where _ spontaneous _ interlayer phase coherence arises .
therefore , the critical layer separation at zero tunneling is of particular interest .
the values of this quantity obtained for finite systems with up to twelve electrons form a rapidly converging data series ( cf figure 4 in ref .
the critical values @xmath37 for the layer separation at zero tunneling are shown in figure [ diagfig1 ] as a function of the ratio @xmath38 of well width to layer separation .
a value of @xmath39 corresponds to the samples used in refs .
@xcite , where the experimental value of @xmath40 for the critical layer separation agrees very well with the exact diagonalization result of @xmath41 . in order to further investigate the order of the quantum phase transition
, we introduce the ratio @xmath42 where the subscript @xmath43 refers to the system size . as discussed in ref .
@xcite , this type of ratio defines a characteristic energy scale of the system at the phase boundary and should prove to be a powerful general tool in the analysis of any quantum phase transition . for a continuous phase transition one would clearly expect @xmath44 to vanish at the phase boundary for an infinite system , while a finite limit @xmath45 is indicative of a finite energy scale , i.e. a first order transition . from our finite - size data for @xmath44 ( evaluated at vanishing tunneling and @xmath46 ) shown in figure [ diagfig2 ]
we conclude that this quantity extrapolates for @xmath47 to a rather substantial non - zero value of order @xmath48k for all values of @xmath34 considered here . along with the arguments and experimental findings given so far
, this result strongly suggests that the bilayer quantum hall system at filling factor @xmath0 undergoes a single first order phase transition as a function of the ratio of layer separation and magnetic length at all values of the tunneling amplitude .
the phase boundary separates a phase with strong interlayer correlation ( and a finite gap for charged excitations ) from a phase with weak interlayer correlations and vanishing gap for charged excitations .
this work was supported by the deutsche forschungsgemeinschaft , the national science foundation , and the welch foundation . as a function of the system size for various ratios of well width @xmath34 to layer separation @xmath50 .
assuming that the these data curves remain of a convex shape also for larger @xmath43 , one concludes that @xmath44 extrapolates to finite values ( being of order 0.05 in units of the coulomb energy scale ) for all values of @xmath38 .
[ diagfig2],width=287 ]
|
we present an effective hamiltonian for a bilayer quantum hall system at filling factor @xmath0 neglecting charge fluctuations .
our model is formulated in terms of spin and pseudospin operators and is an exact representation of the system within the above approximation .
we analyze its low - lying excitations in terms of spin - wave theory .
moreover we add to previous first - principle exact - diagonalization studies concentrating on the quantum phase transition seen in this system .
|
there are about 200 known x - ray binaries in our galaxy .
these contain either a neutron star or a black hole ( @xcite ) .
about one third of them are classified as transients ( see @xcite for a review).some of the x - ray transients , most notably be - binary pulsars , erupt on regular intervals , while others erupt dramatically , but irregularly , with peak luminosities above 10@xmath3 erg / s and typical decay timesorder a month .
this last group represent the majority of known galactic black hole candidates , and provides the best opportunity to study the processes near black hole in relative proximity to the earth .
galactic transients have been therefore a focus of substantial observational and theoretical effort by x - ray astronomers .
much less , however , is known about x - ray transients in other galaxies .
previous observations with _ einstein _ and _ rosat _ were too sparse in time for any systematic monitoring of even nearby galaxies .
the new capabilities provided by _
chandra _ and _ xmm - newton _ started to make the difference .
x - ray observations of the andromeda galaxy ( m31 ) , the closest spiral galaxy to our own ( we assume @xmath4760 kpc distance throughout this letter - van der bergh 2000 ) , are of special interest , because m31 is in many respects similar to the milky way .
m31 was observed previously with _ einstein _ and _ rosat _ , and hundreds of x - ray sources were detected ( @xcite ; @xcite ; @xcite ; @xcite ) .
this letter discusses transients discovered in the m31 core in recent sensitive observations by _ chandra _ and _ xmm - newton_.
we analyzed a series of _ chandra _ data consisting of seven observations with advanced ccd imaging spectrometer ( acis ) and seven high resolution camera ( hrc ) observations .
we used all publicly available observations of m31 central region covering the period from oct 1999 to july 2000 ( table [ obslog ] ) .
a more detailed description of the observations can be found in @xcite ( hereafter g00 ) and @xcite .
the data were processed using the ciao v2.1.2 threads .
we generated x - ray images of the central region of m31 ( fig .
[ image_general ] ) , and performed a wavelet deconvolution of the images to detect and localize point sources . because of calibration uncertainties below 0.3 kev and high background above 7 kev , we excuded acis data outside these energy limits . to estimate energy fluxes and spectra , we extracted counts within source circular regions of @xmath5 to @xmath6 radii ( depending on the distance of the source from the telescope axis ) .
background counts were extracted from the adjacent source - free regions .
spectra of the bright hard sources ( xmmu j004234.1 + 411808 and cxo j004242.0 + 411608 ) were fitted with an absorbed power law model using xspec v.11 .
the acis count rates were converted into energy fluxes using the analytical fits to the spectra for brighter sources or pimms with standard parameters otherwise .
for the hrc the source fluxes were estimated by pimms , based on the observed count rates and the spectral parameters measured by acis and _ xmm - newton_. the central region of m31 was observed with _
xmm - newton _ on june 25 and december 28 , 2000 ( @xcite , hereafter s01 ; @xcite , hereafter o01 ; @xcite , hereafter t01 ) as a part of performance verification ( pv ) program ( table [ obslog ] ) . in our analysis
we used the data from two european photon imaging camera ( epic ) instruments : the mos1 and mos2 detectors ( @xcite ) . in all observations the epic instruments operated in
_ full window _ mode ( @xmath7 fov ) with a medium optical blocking filter .
we reduced epic data with the _ xmm - newton _ science analysis system ( sas v5.01 ) . to obtain the spectra of individual point sources
, we used a source circle of @xmath8 and took backgrounds from adjacent source - free regions .
we used data in the @xmath9 kev energy range , for consistency with _
chandra _ data and because of uncertainties in the calibration of the epic instruments .
this is the default energy band for all fluxes and luminosities presented below .
the energy spectra of the two hard sources ( xmmu j004234.1 + 411808 and cxo j004242.0 + 41608 ) were fitted to an absorbed power law , and an absorbed black body radiation model was applied to fit the spectrum of the supersoft transient xmmu j004319.4 + 411759 ( table [ spec_par ] ) . since the flux from the latter was negligible above 1.5 kev , we considered only its @xmath10 kev spectrum .
epic - mos1 and mos2 data were fit simultaneously , but with independent normalizations .
a new bright source , xmmu j004234.1 + 411808 , was discovered on june 25 , 2000 during the first _ xmm - newton _ observation of m31 ( s01 , o01 , t01 ) .
our analysis of the archival _ chandra _ data revealed the presence of a bright x - ray source at the position consistent with xmmu j004234.1 + 411808 as early as june 21 , 2000 , and again on july 2 and july 29 ( table [ obslog ] ) .
based on the _ chandra _ aspect solution , which is currently limited by systematics to @xmath11 accuracy , we measure the position of xmmu j004234.1 + 411808 to be @xmath12 , @xmath13 ( 2000 equinox ) . the x - ray light curve of xmmu j004234.1 + 411808 shown in fig . [ lc_general]a . combining the results of _ chandra _ and _ xmm _ observations
, we place an upper limit on the effective duration of the outburst of @xmath1440 days .
we have no measurements of the outburst rise , but it was shorter than 20 days and followed by a slower decay .
a fit to decaying part of the light curve with an exponential gives a characteristic e - folding time : @xmath15 days .
the source flux declined from @xmath16 erg s@xmath2 @xmath17 on june 21 to @xmath18 erg s@xmath2 @xmath17 on july 29 , which corresponds to the luminosity drop from @xmath19 erg s@xmath2 to @xmath20 erg s@xmath2 . from the acis observations prior to the outburst , and assuming the same spectrum as measured by _
xmm _ , we compute a @xmath21 upper limit to the quiescent luminosity of @xmath22 erg s@xmath2 , @xmath23 times lower than maximum measured luminosity .
the energy spectrum of the source measured with acis - s on july 2 is similar to one detected with epic - mos a week earlier ( fig .
[ spec_fig]_a _ ) and both may be fit to an absorbed power law with @xmath242.1 ( see table [ spec_par ] for fit parameters ) .
the energy spectrum and x - ray luminosity of xmmu j004234.1 + 411808 correspond to the hard / low spectral state of galactic black hole candidates ( @xcite ) .
moreover , the x - ray evolution of the source is typical for well established class of x - ray novae with fast - rise - exponential - decay ( fred- type ) lightcurves ( @xcite ) . for many of these objects
, radial velocity measurements proved that binary systems harbor black holes .
we propose that xmmu j004234.1 + 411808 is an x - ray nova and the first stellar mass black hole candidate in m31 .
the x - ray source cxo j004242.0 + 411608 was discovered on december 1999 during the first _
observation of m31 ( g00 ) and remained detectable throughout whole y2k ( o01 , t01 ) .
the source was not detected with _
xmm _ on june 29 , 2001 ( @xcite ) , which sets a lower limit of @xmath4400 days to the outburst duration .
the source exhibited remarkable stability of the x - ray luminosity and spectral shape throughout the outburst ( fig .
[ lc_general]_b , c _ and table [ spec_par ] ) .
the average energy spectrum may be approximated with an absorbed power law with @xmath25 .
we note that the acis spectra of the source tend to be generally flatter than epic - mos spectra ( table [ spec_par ] ) .
best - fit values of the absorbing column are also systematically higher for the acis . both discrepancies may be possibly attributed to the presence of a @xmath26 pile - up in the acis spectra .
long ( @xmath271 year ) plateaus were detected in light curves of several black hole candidates , whereas shorter plateaus seem more typical for neutron star systems ( chen et al .
the hard spectrum and long plateau make this outburst remarkably similar to the long - lasting hard transient grs 1716 - 249 ( @xcite ; @xcite ) .
the x - ray transient source xmmu j004319.4 + + 411759 was discovered on june 25 , 2000 by _ xmm _ ( s01,o01 ) .
our analysis of may 26 _ chandra _ data revealed the presence of a bright x - ray source at the position @xmath28 , @xmath29 ( 2000 eq . ,
error circle @xmath30 ) , consistent with previously reported _
xmm _ position ( o01).the source remained active during the june 21 _ chandra _ observation . an energy spectrum obtained with epic- mos1 detector on june 25 is shown in figure [ spec_fig]_d_. it may be satisfactorily fit to an absorbed black body radiation model with a characteristic temperature of @xmath31 ev and absorbing column of @xmath32 @xmath17 .
the spectrum of xmmu j004319.4 + 411759 is similar to the galactic supersoft sources and may be interpreted as a result of thermonuclear burning of the accreted matter on the surface of the white dwarf ( see @xcite for a review ) .
the source proved to be a pulsator with a @xmath33s period ( for more detailed discussion see o01 ) .
the transient behavior of the pulsator hints that it may be a classical nova in the supersoft x - ray spectral phase , several tens of days after the peak of the outburst ( @xcite ; o01 )
. given a number of transient sources detected with _
chandra _ and _ xmm _ , one can estimate the total rate of transient outbursts in the central bulge of m31 . assuming a luminosity of @xmath34 erg s@xmath2 and an average source outburst duration to be more than a month , and taking into account the spacing of the _ chandra _ and _ xmm _ observations
, we estimate a rate of @xmath0 year@xmath2 .
we note a large uncertainty of this estimate , due to the limited statistics . about @xmath35 of a total amount of transients detected in m31
are hard x - ray sources corresponding to @xmath1 outbursts per year .
it is interesting to compare their number with a number of x - ray transient detected in the corresponding region of the galaxy .
based on regular monitoring of the galactic center with _
rxte _ and _ bepposax _ , the rate of hard transient outbursts in this region can be estimated as @xmath36 year@xmath2 , which is comparable to our estimate for the central part of m31 .
this is firther evidence for the similarity between the stellar populations of the galaxy and m31 .
this research has made use of data obtained through the _ chandra _ public data archive .
this paper is based in part on observations obtained with _ xmm - newton _ , an esa science mission with instruments and contributions directly funded by esa member states and the usa ( nasa ) .
chen , w. , shrader , c. r. , livio , m. 1997 , apj , 491 , 312 distefano , r. , kong , a. k. h. , garcia , m. r. , barmby , p. , greiner , j. , murray , s. s. , @xmath37 primini , f. a. 2001 , apj , submitted ( atro - ph/0106254 ) garcia , m. r. , murray , s. s. , primini , f. a. , forman , w. , mcclintock , j. , & jones , c. , 2000 , apj , 537 , l23 ( g00 ) jansen , f. et al .
, 2001 , a&a , 365 , l1 kahabka , p. , & van den heuvel , e. p. j. 1997 , ara@xmath37a , 35 , 69 nousek , j. a. , et al . 1998 , proc .
spie , 3444 , 225 osborne , j. p. , borozdin , k. n. , trudolyubov , s. p. , priedhorsky , w. c. , soria , r. , shirey , r. , guainazzi , m. , hayter , c. , la palombara , n. , mason , k. , molendi , s. , paerels , f. , pietsch , w. , read , a. m. , tiengo , a. , watson , m. g. , @xmath37 west , r. g. 2001 , a@xmath37a , submitted ( o01 ) primini , f. a. , forman , w. , & jones , c. , 1993 , apj , 410 , 615 primini , f. , garcia , m. , murray , s. , forman , w. , jones , c. , @xmath37 mcclintock , j. 2000 , proceeding of the interstellar medium in m31 and m33 , in press ( astro - ph/0012164 ) revnivtsev , m. , gilfanov , m. , churazov , e. , sunyaev , r. , borozdin , k. , alexandrovich , n. , khavenson , n. , chulkov , i. , goldwurm , a. , ballet , j. , denis , m. , laurent , p. , roques , j .-
borrel , v. , bouchet , l. , jourdain , e. 1998 , a&a , 331 , 557 shirey , r. , soria , r. , borozdin , k. , osborne , j. p. , tiengo , a. , guainazzi , m. , hayter , c. , la palombara , n. , mason , k. , molendi , s. , paerels , f. , pietsch , w. , priedhorsky , w. , read , a. m. , watson , m. g. , west , r. g. , 2001_a _ , a@xmath37a , 365 , l195 ( s01 ) shirey , r. 2001 , iau circ 7659 strueder , l. et al . , 2001 ,
a&a , l18 sunyaev , r. a. , borozdin , k. n. , aleksandrovich , n. l. , arefiev , v. a. , kaniovsky , a. s. , efremov , v. v. , maisack , m. , reppin , c. , skinner , j. k. 1994 , astro .
, 20 , 777 supper , r. , hasinger , g. , pietsch , w. , truemper , j. , jain , a. , magnier , e. a. , lewin , w. h. g. , & van paradijs , j. , 1997 , a@xmath37a , 317 , 328 supper , r. , hasinger , g. , lewin , w. h. g. , magnier , e. a. , van paradijs , j. , pietsch , w. , read , a. m. , & truemper , j. 2001 , a@xmath37a , in press tanaka , y. , @xmath37 lewin w. h. g. 1995 , in x - ray binaries , ed
. w. h. g. lewin , j. van paradijs , @xmath37 e. p. j. van den heuvel ( cambridge : cambridge univ .
press ) , 126 tanaka , y. , & shibazaki , n. 1996 , ara@xmath37a , 34 , 607 trinchieri , g. , & fabbiano , g. , 1991 , apj , 382 , 82 trudolyubov , s. et al .
2001 , in preparation turner , m. et al . , 2001 , a&a , 365 , l27 van paradijs , j. 1995 , in x - ray binaries , ed .
w. h. g. lewin , j. van paradijs , @xmath37 e. p. j. van den heuvel ( cambridge : cambridge univ .
press ) , 536 ccccccc + date & photon & n@xmath38&flux@xmath39&l@xmath40&@xmath41 & instrument + ( ut ) & index & ( @xmath42 @xmath17 ) & & & ( d.o.f . ) & + + 13/10/1999 & @xmath43 & @xmath44 & @xmath45 & @xmath46 & 58.2(44 ) & acis - i + 11/12/1999 & @xmath47 & @xmath48 & @xmath49 & @xmath50 & 34.9(42 ) & acis - i + 27/12/1999 & @xmath51 & @xmath52 & @xmath53 & @xmath54 & 48.6(44 ) & acis - i + 29/01/2000 & @xmath55 & @xmath56 & @xmath57 & @xmath58 & 42.8(53 ) & acis - i + 16/02/2000 & @xmath59 & @xmath60 & @xmath61 & @xmath62 & 52.3(62 ) & acis - i + 01/06/2000 & @xmath63 & @xmath64 & @xmath65 & @xmath66 & 45.6(46 ) & acis - s + 25/06/2000 & @xmath67 & @xmath68 & @xmath69 & @xmath70 & 212.2(230 ) & epic - mos + 02/07/2000 & @xmath71 & @xmath72 & @xmath73 & @xmath74 & 38.8(23 ) & acis - s + 28/12/2000 & @xmath75 & @xmath76 & @xmath77 & @xmath78 & 52.3(44 ) & epic - mos + + 25/06/2000 & @xmath79 & @xmath80 & @xmath81 & @xmath82 & 79.9(96 ) & epic - mos + 02/07/2000 & @xmath83 & @xmath84 & @xmath85 & @xmath86 & 4(5 ) & acis - s + + date & t@xmath87&n@xmath38&flux@xmath88&l@xmath89&@xmath41 & instrument + ( ut ) & ( ev ) & ( @xmath42 @xmath17 ) & & & ( d.o.f . ) & + + 25/06/2000 & @xmath90 & @xmath91&@xmath92&@xmath93&53.8(48 ) & epic - mos +
|
we study three transient x - ray sources , that were bright in the central region of m31 galaxy in the year 2000
. observations with _
chandra _ and _ xmm - newton _ allowed us for the first time in the history of x - ray astronomy , to build light curves of transient sources in m31 suitable for studying their variability on a time scale of months and , in some periods , weeks .
the three sources demonstrate distinctly different types of x - ray variability and spectral evolution .
xmmu j004234.1 + 411808 is most likely a black hole candidate based on the similarity of its x - ray light curve and spectra to typical transient low - mass x - ray binaries observed in our galaxy .
the outburst of cxo j004242.0 + 411608 lasted longer than a year , which makes the source an unusual x - ray transient .
the supersoft transient xmmu j004319.4 + 411759 is probably a classical nova - like system containing a magnetized , rapidly - spinning white dwarf .
we estimate a total rate of x - ray transient outbursts in the central bulge of m31 to be of the order @xmath0 per year . the rate of the hard x - ray transients ( @xmath1 year@xmath2 ) in the central part of the andromeda galaxy appears to be comparable to that of the central part of our own galaxy . *
# 1#1
|
the structural transition from a linear chain of repulsively interacting particles to a planar configuration , in the form of a zigzag structure , has been often discussed in theoretical studies on atomic and condensed matter systems .
examples include electrons in nanowires @xcite , ultracold dipolar gases @xcite , vortex lines in bose - einstein condensates @xcite , and ion coulomb crystals in traps @xcite .
specifically in ion coulomb crystals this transition has been experimentally observed and characterized @xcite , thereby determining a phase diagram of the ionic structures as a function of the trap aspect ratio and of the mean interparticle distance @xcite .
theoretical studies demonstrated that the transition from the string to the zigzag chain is associated with a symmetry breaking .
more specifically , if the hamiltonian is invariant under rotation about the string axis , it is the rotational symmetry around the chain which is broken in the zigzag phase .
if instead the motion of the interacting particles is confined to the plane , the zigzag phase breaks the symmetry by reflection about the chain axis .
the phase transition predicted by the classical theory is hence second - order @xcite .
it can be described by a landau model , of which critical values and exponents are well known @xcite , and where the soft mode is the zigzag mode of the linear chain namely , the transverse mode with the shorthest wavelength which drives the instability and determines the new structure .
the corresponding ginzburg - landau equation in the continuum limit and in presence of damping was reported in ref .
, where special focus was given on creation of defects when quenching the value of the transverse trap frequency across the mechanical instability .
further works discussed the structural transitions when the external potential is not harmonic @xcite , showing that the statistical mechanics of the system at the instability may be profoundly modified .
the results of these studies are strictly valid in the classical regime , when the thermal fluctuations can be neglected .
is the temperature of the sample and the dimensionless parameter @xmath2 is tuned by the confining potential or the interparticle distance .
the value @xmath3 corresponds to the value of the transverse frequency , where the linear chain is classically unstable @xcite .
the quantum critical point , at @xmath4 , separates the linear chain from the zigzag phase at @xmath5 . for @xmath6
quantum fluctuations dominate , and the crystal is in the linear ( disordered ) phase .
the dashed lines indicate the boundaries of the quantum critical region , where thermal fluctuations dominate . ] in ref .
we addressed the question , whether quantum fluctuations significantly modify the location of the critical point and the behavior in its vicinity , focussing on ion coulomb crystals .
this question draws on numerical studies in low dimensional dipolar systems , which showed a significant contribution of quantum fluctuations at the transition from a string to a zigzag order @xcite . in this article
we provide the detailed calculations at the basis of the results presented in ref . , and extend the treatment to interacting atomic chains , with repulsive interaction potential scaling with @xmath7 with @xmath8 .
we focus on the case where the motion of the particles is confined to a plane .
as argued below , the string - to - zigzag transition is indeed a quantum phase transition analogous to the ferromagnetic transition of an ising chain in a transverse field @xcite . a phase diagram summarizing the behavior in its vicinity
is depicted in fig .
[ fig:1 ] .
note that a similar mapping to the ising model was argued in the context of electronic systems in quantum wires @xcite .
however , the validity of the assumed equivalence between a continuous field theoretical model and a discrete ( @xmath9 ) field - theory can not be justified on the basis of classification into universality classes below three space dimensions @xcite . in the present study
we develop a quantum field theoretical approach , which allows a direct mapping to the quantum one dimensional ising model under plausible assumptions . on this basis
we determine the quantum critical point at which the transition occurs and identify the experimental parameters , for which it can be distinguished from the value predicted by the classical theory @xcite .
in particular , we relate the critical value of the transverse trap frequency @xmath10 to the ratio of the typical kinetic energy scale @xmath11 and interaction energy @xmath12 . defining a parameter @xmath13 , which describes the deviation of the transverse confinement from its stability point
, we find that the critical point is given by @xmath14 .
a measurement is suggested , which can demonstrate transverse quantum fluctuations in a long chain of trapped ions ( composed of hundreds of ions ) and probe the different regimes in the phase diagram .
this paper is organized as follows . in sec .
[ sec : model ] we derive a quantum field theory for the zigzag phonon mode .
the mapping to the quantum ising model is described in sec .
[ sec : mapping ] . in sec .
[ sec : qcp ] we derive expressions for the ising model parameters , and in particular their critical values , in terms of microscopic parameters of the interacting atoms system . in sec .
[ sec : experiment ] we discuss an experimental realization in trapped ions systems , and specify the conditions under which the quantum phase transition is likely to be observable .
finally , our conclusions and outlook are summarized in sec . [
sec : conclude ] .
we start by considering a linear array of particles with mass @xmath15 , which are confined in the @xmath16 plane .
the system is at low temperature , and the particles at equilibrium are aligned along the @xmath17-axis and equidistant , with interparticle distance @xmath18 such that their equilibrium position is @xmath19 and @xmath20 , while the motion along the @xmath21-axis is assumed to be frozen out .
periodic boundary conditions are assumed at the chain edges and one ion of the chain is assumed to be pinned .
the particles can oscillate around the equilibrium positions in the @xmath17 and @xmath22 direction , where they are confined by a harmonic potential of the form @xmath23 here @xmath24 is the displacement of the @xmath25th atom from the chain axis and @xmath10 the trap frequency .
finally , they interact by means of the repulsive power law potential , @xmath26^{\alpha/2 } } \label{vint}\ ] ] where @xmath27 denotes the atomic position in the plane , @xmath28 is the strength of the interaction and @xmath8 .
this model includes the coulomb interaction when @xmath29 and the dipolar interaction when @xmath30 , assuming that the dipoles are polarized along the @xmath21 axis , such that in the @xmath16 plane they experience an isotropic repulsive force .
we also note that in the case of the coulomb interaction the strength is such that @xmath31 , where @xmath32 is the charge of the particle . in this case
the array can result from the self - organization of the atoms , since quasi - long - range order can be assumed as long as one particle in the chain is pinned . the configuration , in which the particles are aligned along the @xmath17-axis with uniform interparticle distance , could be realized in a ring trap at ultralow temperatures under the condition that one ion is pinned @xcite , or in a anharmonic linear paul trap @xcite .
it also corresponds with good approximation to the ions distribution at the center of a long chain in a linear paul trap @xcite . for @xmath33 ,
on the other hand , particle - particle correlations decay at @xmath5 with a power law dependence on the distance @xcite , and long - range order may be only assumed in presence of an additional periodic potential along @xmath17 , which localizes the particles at the positions @xmath34 . for dipolar atoms or molecules with an electric dipole moment @xmath35 , this configuration
could be realized by means of an optical lattice @xcite with periodicity @xmath18 . here
, the dipole - dipole interaction can be made repulsive in the @xmath16 plane when a static electric field along the @xmath21 axis aligns the dipole .
then , the strength of the interaction is such that @xmath36 .
we assume temperatures such that the atoms crystallize and their motion is well approximated by harmonic vibrations about the equilibrium positions when the configuration is mechanically stable . while the linear chain is stable against longitudinal displacements from the equilibrium positions , a soft mode in the transverse direction drives a structural phase transition of the linear chain into a zigzag configuration , as shown in ref . .
this mode has wavelength @xmath37 , such that in the classical harmonic crystal the corresponding displacements are described by the function @xmath38 with @xmath39 the amplitude of the oscillations .
we therefore focus on the dynamics of the transverse phonon modes close to the instability and expand @xmath40 to fourth order in @xmath41 .
moreover , we introduce the field @xmath42 , which describes a transverse excitation close to the instability and is slowly varying with respect to the length scale @xmath18 , i.e. @xmath43 using a gradient expansion to leading order @xcite , we obtain the effective potential @xmath44=v_t+v_{int } \approx \frac{1}{2}m\nu_t^2a^2\sum_{j=1}^n \phi_j^2 + \frac{a^2}{2a^\alpha}\sum_{i\not = j}\frac{\alpha}{2}\left\{-\frac{(\phi_i-(-1)^{i - j}\phi_j)^2}{|i - j|^{\alpha+2 } } + \frac{(\alpha+2)(\phi_i-(-1)^{i - j}\phi_j)^4}{4|i - j|^{\alpha+4}}\right\}\label{v_t_tot } \ ; .\end{aligned}\ ] ] terms describing the coupling between the axial and transverse modes have not been reported , because they give rise to higher order corrections in the gradient expansion .
the latter is defined by the small parameter @xmath45 , with @xmath46 the typical deviation between the wave vector @xmath47 of the excited mode and the wave vector @xmath48 of the soft mode .
an extensive discussion on the derivation of the effective potential for the soft mode can be found in ref . .
the ordered , zigzag phase is established when @xmath42 acquires a finite , constant expectation value , as we show below .
to leading order in a gradient expansion , the potential in eq .
( [ v_t_tot ] ) can be mapped to the dynamics described by a potential of the form @xmath49\approx \sum_{j=1}^{n } v_0(\phi_j ) + \frac{1}{2}k\sum_{j=1}^n ( \phi_j-\phi_{j+1})^2\ ; , \label{v_phi}\ ] ] where the interaction is now nearest neighbors . here
@xmath50 is a local potential @xmath51 and the parameters @xmath52 , @xmath53 and @xmath54 are given by @xmath55\alpha a^2}{a^\alpha}\ ; , \quad \nu_c^2=\frac{4c_2[\alpha+2]\alpha a^2}{ma^{\alpha+2 } } \ ; , \nonumber\\ g&= & \frac{8c_2[\alpha+4]\alpha(\alpha+2 ) a^2}{a^{\alpha+4}}\ ; , \label{v_parameters}\end{aligned}\ ] ] with @xmath56\equiv \sum_{j=1}^n \frac{(-)^{j+1}}{j^\alpha}\ ; , \quad c_2[\alpha]\equiv \sum_{j\geq 1\ , , odd } \frac{1}{j^\alpha}\ ; .\nonumber\end{aligned}\ ] ] a classical theory , which neglects quantum fluctuations ( i.e. , valid in the limit @xmath57 ) , predicts a transition from a linear to a zigzag chain at @xmath5 when the transverse confining frequency @xmath10 is reduced below the critical value @xmath53 @xcite .
for later convenience we define a dimensionless parameter describing the deviation of the transverse trap frequency from this classical transition point : @xmath58 for @xmath59 , the local potential @xmath50 [ eq .
( [ v_loc ] ) ] has minima at @xmath60 where the frequency @xmath61 is given by the relation @xmath62 and is associated with the curvature of @xmath50 at the top of the barrier between the minima of the double well potential , as one can observe from eq .
( [ v_loc ] ) .
we note that the mapping to eq .
( [ v_phi ] ) is valid for a power - law interaction with @xmath8 ( for which the relevant sums @xmath63 $ ] in eq .
( [ v_parameters ] ) converge ) , hence including the case of the coulomb interaction at @xmath29 .
therefore , for the linear - zigzag instability the coulomb interaction effectively belongs to the class of short - range potentials .
indeed , one finds that the spectrum of the transverse excitations close to the instability ( approaching from the side where the chain is stable classically ) behaves as @xmath64 for @xmath8 ( see for instance ref .
for the case @xmath29 ) .
this behavior is markedly different from that of the long wavelength , axial excitations : for @xmath33 the axial frequency @xmath65 , with @xmath66 sound velocity , while for @xmath29 there exists no sound velocity , and @xmath67 @xcite .
these considerations on the structure , which are based on identifying the minima of the potential energy , do not account for the contribution of the kinetic energy .
thermal effects are expected to modify the behaviour at the critical point . moreover , even at temperatures @xmath68 , when thermal fluctuations are small , quantum fluctuations will become relevant . sufficiently close to the value @xmath53 , i.e. for sufficiently small values of the parameter @xmath2 ( where the classical zigzag ordering is expected ) , quantum fluctuations will induce tunneling between degenerate minima of the potential and are expected to destroy the zigzag ordering .
a true phase transition will therefore occur at a smaller value of @xmath10 , corresponding to a _
quantum _ critical point , such that @xmath69 ; i.e. , at @xmath70 . in order to investigate the quantum critical behavior , one has to introduce a 1 + 1 dimensional quantum field
theory for the system . to this end
, we write the partition function as @xmath71/\hbar } \label{z_def}\ ] ] with the euclidean action @xmath72&= & \int_0^{\hbar\beta } d\tau \sum_{j=1}^n \left[\frac{1}{2}ma^2(\partial_\tau \phi_j)^2\right . \nonumber\\ & + & v_0(\phi_j)+\left.\frac{1}{2}k(\phi_j-\phi_{j+1})^2\right]\ ; , \label{s_phonons}\end{aligned}\ ] ] where @xmath73 and @xmath74 for @xmath68 .
below we demonstrate the mapping of this model to the ising chain in a transverse field , and derive approximate expressions for its parameters in terms of microscopic parameters of the interacting particles system .
this will facilitate the study of the conditions under which the phase transition can be observable in such systems .
the low energy model derived in the previous section [ eqs .
( [ z_def ] ) , ( [ s_phonons ] ) ] describes the quantum dynamics of the zigzag phonon mode in terms of a real continuous scalar field @xmath75 .
we now express it in a form which will allow its mapping onto an effective model for the discrete field @xmath76 $ ] . as a starting point
, we perform the standard procedure of dividing the imaginary time ( @xmath77 ) axis into discrete time steps @xmath78 separated by an infinitesimal interval of size @xmath79 , with @xmath80 and @xmath81 .
the partition function is then cast in the form @xmath82 where @xmath83 is a transfer matrix describing the time evolution from @xmath84 to @xmath85 . using the notation @xmath86 , @xmath87 ,
the matrix elements of @xmath83 are given by @xmath88&=&\exp\left\{-\frac{\delta\tau}{2\hbar } \sum_{j=1}^n\left[ma^2\left(\frac{\phi_j-\phi^\prime_j}{\delta\tau}\right)^2 + v_0(\phi_j)+v_0(\phi^\prime_j ) + \frac{1}{2}k\left((\phi_j-\phi_{j+1})^2+(\phi^\prime_j-\phi^\prime_{j+1})^2\right)\right]\right\}\;\\ & = & \prod_{j=1}^ng(\phi_j,\phi^\prime_j;\delta\tau)\;\exp\left\{-\frac{\delta\tau}{4\hbar}k \sum_{j=1}^n\left[(\phi_j-\phi_{j+1})^2+(\phi^\prime_j-\phi^\prime_{j+1})^2\right]\right\}\ ; .
\label{t_loc}\end{aligned}\ ] ] here the propagator @xmath89 is given by @xmath90 in which the local hamiltonian @xmath91 describes the quantum dynamics of a particle in the double
well potential @xmath92 , where @xmath93 is the momentum conjugate to @xmath42 .
the symmetry of the potential implies that the exact eigenstates of @xmath94 consist of pairs of states with well - defined symmetry under @xmath95 : symmetric ( @xmath96 ) and antisymmetric ( @xmath97 ) .
the corresponding energy eigenvalues can be expressed as @xmath98 , @xmath99 respectively , where the splitting energy @xmath100 is associated with the overlap of wave functions centered in either of the two wells . the propagator @xmath101 ( the index @xmath25 is dropped to simplify the notation ) therefore acquires the form @xmath102 we now change basis into wave -functions centered at the right and left potential minima @xmath103 [ eq . ( [ phi_0 ] ) ] : @xmath104 equation ( [ prop_sa ] ) becomes @xmath105 \cosh\left\{\frac{\delta \omega_n\delta\tau}{2}\right\}\right.\nonumber\\&+&\left[\langle \phi^\prime |r_n\rangle \langle
l_n|\phi\rangle + \langle \phi^\prime |l_n\rangle \langle r_n|\phi\rangle\right]\left .
\sinh\left\{\frac{\delta \omega_n\delta\tau}{2}\right\}\right\}\ ; . \nonumber\end{aligned}\ ] ] we next define the wave functions @xmath106 , where @xmath107 denotes the isospin @xmath108 , respectively . since these wave functions are not known exactly , we implement a variational approach and assume the trial function @xmath109 where @xmath110 for @xmath111 is an oscillatory , symmetric function of @xmath112 with a number of nodes increasing with @xmath113 ; for the lowest energy states @xmath114 we assume @xmath115 , so that @xmath116 is a single variational parameter . the dependence of @xmath117 on the magnitude and sign of @xmath112 can be made explicit using the substitution @xmath118 , which yields @xmath119 the first term in eq .
( [ wf_sigma ] ) dominates as long as @xmath120 . to complete the derivation of an effective field theory in terms of the discrete fields @xmath121
, one needs to perform the path integral [ eq .
( [ z_def ] ) ] over the magnitude - field @xmath122 .
due to the gaussian factor in eq .
( [ wf_sigma ] ) , the integration over @xmath123 ( for each @xmath113 ) can be written schematically as @xmath124 where @xmath125 encodes the remaining @xmath42-dependence , including in particular the interparticle coupling terms @xmath126 [ the last term in eq .
( [ s_phonons ] ) ] .
if we now consider the case where @xmath127 ( which is valid deep in the zigzag phase ) , a saddle - point approximation of the integral yields ( up to a multiplicative constant factor ) @xmath128 .
this implies that the partition function [ eq . ( [ z_t ] ) ] can be recast as @xmath129 with @xmath130 and @xmath131 a @xmath132matrix related to the original transfer matrix @xmath83 by @xmath133=t[\{\phi_0\sigma_j\},\{\phi_0\sigma^\prime_j\}]\ ; . \label{teff_def}\ ] ] recalling eq .
( [ t_mat ] ) , we find that @xmath134 $ ] can be expressed as @xmath135 \label{teff}\\&=&\prod_{j=1}^n t_{loc}[\sigma_j,\sigma^\prime_j]\exp\left\{\frac{k\phi_0 ^ 2\delta\tau}{2\hbar}\left(\sigma_j\sigma_{j+1}+\sigma^\prime_j\sigma^\prime_{j+1}\right ) \right\}\nonumber \;\end{aligned}\ ] ] where @xmath136=const\times g(\phi_0\sigma,\phi_0\sigma^\prime;\delta\tau)$ ] , and the propagator @xmath137 can be obtained by inserting eq .
( [ wf_sigma ] ) in eq .
( [ prop_rl ] ) . the final stage in the mapping to the quantum ising chain involves a derivation of an explicit expression for the @xmath138matrix @xmath139 , which encodes the single particle dynamics .
by inspection of eqs .
( [ wf_sigma ] ) and ( [ prop_rl ] ) it is evident that @xmath139 obeys the symmetry @xmath136=t_{loc}[-\sigma,-\sigma^\prime]$ ] .
it can therefore be written in the form @xmath140 where @xmath141 are pauli matrices in the basis where @xmath142 denote the eigenvalues of @xmath143 , and @xmath144 is the @xmath138 unit matrix .
this is the crucial point where we formulate the symmetry of the problem , which stems from the symmetry relations leading to eqs .
( [ prop_def])-([rl_def ] ) .
the general expressions for @xmath145 and @xmath146 are @xmath147 where @xmath148 equation
( [ aandb ] ) implies that in general @xmath145 and @xmath146 are non trivial functions of the time interval @xmath79 .
restrictions on the parameters must therefore be set in order to establish a well defined effective model for arbitrarily small @xmath79 .
a simplification can be achieved in view of the fact that the sum over @xmath113 is rapidly converging : a semiclassical calculation indicates that @xmath149 for large @xmath113 ( see appendix for details ) , i.e. the level spacing increases with @xmath113 . in the regime where the tunnel splitting obeys @xmath150 ( realized far enough from the quantum transition so that the wells of @xmath50 are well separated ) , the sum is dominated by the first term @xmath114 up to exponentially small corrections . even closer to the phase transition , where the tunnel splitting becomes comparable to the barrier between the wells , the sum in eq .
( [ aandb ] ) is dominated by the first few terms . as we show in the next section ,
provided the quartic term in @xmath50 ( i.e. , the parameter @xmath54 ) is sufficiently large , all the energy levels @xmath151 are given ( up to a numerical factor ) by _ the same _ energy scale @xmath152 . since the differences @xmath153 are typically smaller than @xmath154 , the dependence on @xmath113 of all terms in the brackets in eq .
( [ aandb ] ) may be neglected .
one can therefore approximate @xmath155 for all @xmath113 by @xmath156 .
this yields @xmath157\ ; . \label{tloc_n0}\ ] ] assuming in addition that the overlap factor @xmath158 [ eq . ( [ overlap ] ) ] obeys @xmath159 eq .
( [ tloc_n0 ] ) reduces to @xmath160 \label{tloc}\\ & \approx & t_0\left(\cosh\left\{\frac{\delta\omega\delta\tau}{2}\right\}\delta_{\sigma\sigma^\prime } + \sinh\left\{\frac{\delta\omega\delta\tau}{2}\right\}\delta_{\sigma,-\sigma^\prime}\right ) \nonumber\end{aligned}\ ] ] where @xmath161 is a constant prefactor . in terms of pauli
s matrices this becomes @xmath162 as a result , the expression for @xmath163 [ eqs .
( [ z_teff ] ) and ( [ teff ] ) ] effectively reduces to @xmath164/\hbar } \label{z_ising}\ ] ] in which @xmath165 is a constant and @xmath166 is the action of an ising chain in a transverse field @xcite subject to the hamiltonian @xmath167 here , the fictitious exchange coupling @xmath168 and transverse field @xmath169 are given by @xmath170 where we have used eq .
( [ phi_0 ] ) for @xmath171 . the model described by hamiltonian ( [ h_ising ] )
is known to exhibit a quantum phase transition at @xmath172 and @xmath5 , separating an ordered phase at @xmath173 ( in our case , the stable zigzag configuration ) from a disordered phase at @xmath174 ( the linear chain ) . in both phases the spectrum of excitations
is characterized by a gap with energy @xmath175 .
the corresponding phase diagram , showing the critical behavior as a function of @xmath2 ( and hence @xmath10 ) and @xmath1 is sketched in fig .
[ fig:1 ] . the critical regime ( @xmath176 )
is characterized by universal power - law @xmath1-dependence of correlation functions with the critical exponents of the quantum transition @xcite . in the next section
we evaluate the gap in terms of the atom chain system parameters , and in particular relate it to the parameter @xmath2 [ eq .
( [ def_epsilon ] ) ] .
this will enable an estimate of the critical value @xmath177 satisfying @xmath178 .
it should be noted that the condition ( [ condition ] ) was required to obtain the form ( [ tloc_sigma_x ] ) for the local transfer matrix , and thus the final stage in the derivation of eqs .
( [ z_ising]),([h_ising ] ) .
namely , the mapping to the ising model eq .
( [ h_ising ] ) is valid only as long as the lowest energy levels are below the barrier in the potential @xmath50 . in this regime ,
the tunneling between the wells is small and hence necessarily @xmath179 .
this is satisfied deep in the ordered ( zigzag ) phase , in which case all our approximations are justified .
it does _ not _ hold near the critical point , which we estimate in the next section . however , we note that conformal symmetry strongly restricts the number of universality classes in @xmath180-d @xcite .
moreover , it was argued that the number of relevant operators in two dimensions is the same as in @xmath181 dimension @xcite , therefore the resulting @xmath0 field theory describing our model is expected to be in the same universality class as the two dimensional ising model .
based on the symmetry properties of the present system ( as manifested by eq .
( [ tloc_gen ] ) and the following derivation ) it is reasonable to assume that a quantum ising model is the appropriate field theory , which correctly describes its critical behavior at @xmath182 .
this hypothesis is further supported by numerical studies , e.g. , by barma and fisher @xcite and a later work by kim , lin and rieger @xcite . under these assumptions
we estimate the quantum critical point in the regime of parameters where classically zigzag order is suppressed by quantum tunneling , and find that it belongs to a universality class @xcite that differs from the one of the classical landau model .
as we show below , the effective coupling parameter @xmath183 of the ising model [ eq .
( [ ising_jh ] ) ] can be expressed in terms of two dimensionless parameters characterizing the system : the first is the ratio of energy scales @xmath184 , where @xmath185 is a typical kinetic energy scale of atoms in the chain , and @xmath186 is an energy scale associated with the interaction . in atomic systems , this ratio can be tuned most effectively by controlling the density of atoms @xmath187 , e.g. , by means of the lateral confinement .
the second dimensionless parameter @xmath2 [ eq .
( [ def_epsilon ] ) ] can in principle be tuned independently by controlling the transverse confining potential frequency @xmath10 .
the expression for the fictitious exchange coupling @xmath168 in terms of the above definitions can be obtained directly from eq .
( [ ising_jh ] ) . employing eqs .
( [ v_parameters ] ) , ( [ omega_def ] ) and ( [ u_p ] ) , we rewrite the parameters as @xmath188\alpha u_p\ ; , \quad m\omega^2=\frac{4c_2[\alpha+2]\alpha u_p\,\epsilon}{a^2 } \ ; , \nonumber \\ g & = & \frac{8c_2[\alpha+4]\alpha(\alpha+2)u_p}{a^4}\ ; . \label{j_dimpar}\end{aligned}\ ] ] this yields @xmath189c_2[\alpha+2]\alpha}{2c_2[\alpha+4](\alpha+2)}\ ; . \label{j_up}\ ] ] note that since we focus on the regime @xmath190 , and by definition @xmath191 , the energy scale @xmath168 can be tuned continuously from @xmath192 to @xmath193 .
a quantum phase transition is expected to occur at @xmath177 , dictated by the ratio @xmath184 at which @xmath168 is equal to the transverse field " energy scale @xmath169 . to get a concrete estimate of @xmath169 in terms of @xmath11 and @xmath12
, we perform a variational calculation of the splitting energy @xmath194 . to this end
, we consider the trial wave functions @xmath195 for the lowest energy states [ i.e. , the symmetric and antisymmetric combinations of eq .
( [ rl_wf ] ) with @xmath114 ] , and search for a minimum of the energies @xmath197 with respect to the variational parameter @xmath198 .
here @xmath199 is the hamiltonian of a particle of mass @xmath15 subject to the double well potential @xmath50 , and @xmath200 .
it is convenient to eliminate the parameters of the quadratic part of @xmath199 by introducing the length scale @xmath201 and the energy scale @xmath202 . using the normalized coordinate @xmath203 we write the hamiltonian
as @xmath204 depending on a single dimensionless parameter @xmath205 the nature of the eigenstates and eigenvalues of @xmath199 crucially depends on @xmath206 being small or large . the exact minimum condition for @xmath207 $ ] yields a cumbersome , transcendental equation for the variational parameter @xmath198 . in particular
, it includes the exponential factors @xmath208 , with @xmath209 , @xmath210 .
these factors are associated with the overlap of the right and left - centered wave functions . however , since we are interested in a regime of parameters close to the critical point ( see discussion between eqs .
( [ h_ukup ] ) and ( [ qcp ] ) ) , we focus on the approximation of a quasi flat " potential , corresponding to @xmath211 and @xmath212 .
essentially , this regime amounts to a dominance of the quartic term in the potential , while the barrier separating the two wells is relatively small .
in fact , the parameter @xmath206 can be written as @xmath213 , where @xmath214 is the difference between the energy of the minima of @xmath50 and the top of the bump separating them . therefore , @xmath206 is the ratio between the zero - point energy @xmath202 and the potential energy of the barrier .
as becomes clear from the following discussion , the regime @xmath215 is relevant for the critical point . in this regime also
the two gaussian functions of eq .
( [ pm_wf ] ) are expected to mix , therefore @xmath216 . to find an analytic expression we take these inequalities to the extreme .
we find from eq .
( [ phi_0 ] ) @xmath217 . to leading order in @xmath218 ,
this yields a minimum of @xmath207 $ ] at @xmath219 both @xmath207 $ ] and @xmath220 $ ] are computed using eq .
( [ epm ] ) within the above approximations , with the value of @xmath221 which was found minimizing @xmath207 $ ] .
we checked that would we extremize @xmath220 $ ] , a value with the same dependence on parameters that differs only by 3% would be obtained .
the resulting energy eigenvalues @xmath222 $ ] are separated by a splitting energy @xmath223-e_+ [ l]$ ] where @xmath224 this sets the scale of the transverse field .
recalling eq .
( [ b_def ] ) for @xmath206 , we note that the dependence on @xmath61 ( and hence on @xmath2 ) cancels out .
we thus get @xmath225\alpha(\alpha+2)}{2}\right)^{1/3 } \label{h_ukup}\end{aligned}\ ] ] where we have used eq .
( [ j_dimpar ] ) to relate @xmath54 to @xmath12 .
it should be pointed out that ( up to a numerical factor close to unity ) the above estimate for @xmath169 coincides with the energy scale dictating the higher energy eigenvalues of @xmath199 in the semiclassical approximation ( formally valid for @xmath226 ) : @xmath227 [ see appendix , eq .
( [ sen2 ] ) ] .
we hence conclude that in this regime of parameters , all energy levels are determined by the same energy scale .
we finally employ eqs .
( [ j_up ] ) and ( [ h_ukup ] ) to evaluate the critical value of @xmath2 for a given @xmath184 : the condition @xmath228 dictates @xmath229 where we have used the approximation @xmath230\sim c_2[\alpha]\sim 1 $ ] for arbitrary @xmath231 . for @xmath29 ,
@xmath30 the numerical prefactor is @xmath232 .
this implies that for given @xmath233 , a quantum critical point is typically expected at @xmath234 .
an interesting system to study the above predicted quantum phase transition would be an ion crystal in a quasi one dimensional trap @xcite . for singly - ionized alkali - earth metal atoms , the generalized charge
@xmath235 is the electron charge ( @xmath236 ) , and the mass of the ions can be written as @xmath237 where @xmath238 is the proton mass and @xmath239 the atomic number . in these systems it is possible to tune through the critical point by either controlling the transverse confinement frequency @xmath10 ( which is typically in the mhz regime ) , or the spacing between neighboring ions @xmath18 which is conveniently measured in units of micrometers : @xmath240 m . to be able to distinguish between the quantum disordered phase and the ordered ( zigzag ) phase
, one must first make sure that the frequency difference @xmath241 implied by our estimated @xmath177 [ via eq .
( [ def_epsilon ] ) ] is not limited by the experimentally accessible resolution . to this end , it is useful to write @xmath177 in terms of @xmath239 and @xmath242 defined above . inserting eqs .
( [ u_k ] ) , ( [ u_p ] ) and the numerical values of @xmath243 , @xmath244 and @xmath238 in eq .
( [ qcp ] ) , we obtain @xmath245 from eq .
( [ def_epsilon ] ) , we therefore obtain an upper bound to the frequency resolution @xmath246 which ranges from 100@xmath247hz ( for protons with @xmath248 ) to 1@xmath247hz for heavier ions . this bound has to be compared with the time scale of heating in ion traps . in order to be able to measure the quantum phase transition , the heating and decoherence time scale of the trap
, @xmath249 should be such that @xmath250 , which leads to demanding conditions for the existing trapping setups , see for instance the discussion in ref . .
another requirement on the experimental setup is the possibility to reduce the temperature @xmath1 below the energy scale characteristic of the gap @xmath251 .
( [ h_ukup ] ) , the estimated restriction on @xmath1 ( defined for convenience in units of milli - kelvin ) is therefore given by @xmath252\ll \frac{10^{3}c_h(u_k^2u_p)^{1/3}}{k_b}\ ; .
\label{t_ukup}\ ] ] to derive a numerical estimate , we again employ eqs .
( [ u_k ] ) , ( [ u_p ] ) and rewrite the mass and inter
ion spacing in terms of @xmath239 , @xmath242 .
this yields @xmath252\ll 0.25\left(\frac{1}{n_a^2a_0 ^ 5}\right)^{1/3}\ ; , \label{t_num}\ ] ] implying an upper bound of order @xmath253mk for protons to @xmath254k for magnesium ions .
these values are challenging for large crystals , but could be accessed when , for instance , only the transverse motion is cooled ( methods for cooling ion coulomb crystals to ultralow temperatures are discussed in ref . ) .
we note that the condition on @xmath255 can be relaxed in the presence of screening with permittivity @xmath256 , with @xmath257 the vacuum permittivity : in this case the potential energy @xmath12 decreases as @xmath258 , and the upper bound on @xmath255 increases with @xmath259 .
it is interesting to note that , correspondingly , the condition of the temperature becomes more restrictive , as the upper bound decreases with @xmath260 .
this situation could possibly be realized , for instance , in the case of an ion chain embedded in a crystal , as in the experiment reported in ref . .
another physical system , where the quantum phase transition can be experimentally observed , are ultracold dipolar gases , such as polar molecules @xcite . here
, the interaction is repulsive provided the dipoles are aligned perpendicularly to the plane where their motion is studied , for instance , in a very steep trap which freezes out the motion in the perpendicular direction . in ref .
a zigzag ordering across tubes , containing gases of ultracold polar molecules , has been predicted when the distance between the tubes was below a certain value .
long - range order is found in the presence of an additional external periodic potential which localizes the dipoles in an equidistant array in one dimension and could be realized by means of an optical lattice .
the regime here studied would then correspond to the realization of a mott - insulator state of the polar molecules in a one dimensional optical lattice , where the transverse direction must be confined by a dipole trap , whose steepness is changed in order to reach several points of the phase diagram @xcite .
an estimate of the parameters , which allow to access the low dimensional , quasi - crystalline region , has been presented in ref . .
as a function of @xmath261 ( in units of the transverse length @xmath262 ) : ( a ) for an ion string ( @xmath263 ) .
the red dashed curve corresponds to the classically stable regime ( i ) , where @xmath264 is given by @xmath265 [ eq . (
[ s : linear ] ) ] ; the blue solid curve corresponds to the ion string in the quantum disordered phase regime ( ii ) ( @xmath266 in eq .
( [ s : disorder ] ) with @xmath267 ) .
( b ) for a zigzag structure , regime ( iii ) ( with @xmath268 ) .
the dashed and solid lines correspond to @xmath269 [ eq . ( [ s : zigzag ] ) ] when @xmath270 with @xmath113 even and odd , respectively.,title="fig:",width=302 ] as a function of @xmath261 ( in units of the transverse length @xmath262 ) : ( a ) for an ion string ( @xmath263 ) .
the red dashed curve corresponds to the classically stable regime ( i ) , where @xmath264 is given by @xmath265 [ eq . (
[ s : linear ] ) ] ; the blue solid curve corresponds to the ion string in the quantum disordered phase regime ( ii ) ( @xmath266 in eq .
( [ s : disorder ] ) with @xmath267 ) .
( b ) for a zigzag structure , regime ( iii ) ( with @xmath268 ) .
the dashed and solid lines correspond to @xmath269 [ eq . ( [ s : zigzag ] ) ] when @xmath270 with @xmath113 even and odd , respectively.,title="fig:",width=302 ] in atomic and molecular systems , the measurement of the disordered phase , where quantum fluctuations dominate in the transverse direction , can be performed by means of the structure form factor , here defined by the expression @xmath271 where @xmath272 is the density of atoms in the two dimensional plane . for wave vectors in the limit @xmath273 , where @xmath48 is the wave number of the zigzag , @xmath274 in which @xmath275 is proportional to the squared order parameter , @xmath276 , while @xmath277 is the contribution of the fluctuations and is proportional to the isothermal susceptibility @xmath278 of the corresponding ising system @xcite .
approaching the quantum critical point for @xmath279 ( i.e. , for @xmath280 below the dashed lines in fig .
1 ) , @xmath281 ) , where @xmath282 and @xmath283 are the exponents of the classical two dimensional ising model . in the critical region @xmath284 ( shaded area in fig .
1 ) , these behave as @xmath285 , @xmath286 respectively , where we have used the critical scaling of the correlation length @xmath287 that is sufficiently different from @xmath288 ) with the help of light scattering at the appropriate wave vector . the structure factor will take a different form in each of three cases : _ ( i ) _ the linear ( string ) chain ( @xmath289 ) , _ ( ii ) _ the quantum disordered phase ( @xmath6 ) , and _ (
iii ) _ the zigzag configuration ( @xmath290 ) for @xmath291 atoms . in regime _
( i ) _ , where the ion string is assumed , the atomic density can be written as @xmath292 such that @xmath293 is the sum of well spatially - separated gaussians , centered in @xmath294 and with width @xmath295 , while @xmath296 is a gaussian with width @xmath297 such that @xmath298 .
the structure form factor is then given by @xmath299 with @xmath300 and corresponds to @xmath301-peaks along the @xmath17-axis at the location of the vectors of the reciprocal lattice , @xmath302 , where the height of the peaks is embedded in a gaussian function of width @xmath303 , while the dependence on @xmath261 has the form of a gaussian function of width @xmath304 ( see fig . [ fig:2](a ) ) .
we next consider the zigzag ordered phase regime _
( iii)_. denoting by @xmath305 the transverse displacement from the @xmath17-axis , for @xmath306 we obtain @xmath307 which provides an interference pattern in the @xmath22-direction , with a maximum or a zero at @xmath308 depending on whether the corresponding @xmath17 component is even or odd , as displayed in fig . [ fig:2](b ) .
we note that the appearance of an interference pattern for @xmath113 _ odd _
( i.e. , the doubling of the unit cell ) is a primary signature of this ordered phase . in both cases ,
the pattern is characterized by peaks separated by @xmath309 .
these expressions are in agreement with the results found numerically for the linear and zigzag configuration in ultracold dipolar gases @xcite . finally , we focus on regime _
( ii ) _ , namely the phase where the zigzag order found in the framework of the classical theory is destroyed as a result of quantum fluctuations , and the disordered linear chain is found .
this regime is characterized by a wave function whose width in the transverse direction is of the order of the distance between the two minima of the potential @xmath50 of eq .
( [ v_loc ] ) , see for instance the numerical result reported in ref . .
for this purpose , we assume that the density along @xmath310 is given by @xmath311 with @xmath312 given in eq .
( [ pm_wf ] ) and @xmath313 giving the right normalization , where @xmath314 is the overlap factor [ eq .
( [ overlap ] ) ] .
the corresponding structure form factor reads @xmath315 where non - locality is here in the dependence on the parameter @xmath158 , giving the overlap between the two wavefunctions in the two wells .
this result was already obtained for the case of a single atom in a double well potential @xcite , and can be understood as an interference pattern arising from a photon , which is elastically scattered by one of the atoms . in the disordered phase
the scattered photon will follow two possible pathways , associated with the the coherent superposition of the two positions where the atom can be found with largest probability .
figure [ fig:2](a ) displays the function @xmath266 as a function of @xmath261 . the existence of sidebands ( fig .
[ fig:2](a ) ) is the signature of the regime _
( ii)_. this demonstrates how one can distinguish the various regimes and measure the shift of the critical point resulting from quantum tunnelling .
we also remark that autocorrelation functions of the crystal can be measured by means of ramsey interferometry , by driving the internal transition of one ion in the chain , as proposed in ref . .
this method would allow one to extract the critical exponents characterizing the critical behavior .
methods for revealing quantum tunneling between zigzag configurations of few ions have been proposed in ref . .
in this paper , it has been argued that the structural phase transition between a chain and a zigzag for ions or atoms interacting via a power - law interaction falling off as @xmath316 ( where @xmath317 is the distance and @xmath318 ) can be mapped on an ising model in a transverse field . in this way the classical theory for the phase transition can be extended to account for quantum fluctuations .
such correspondence was proposed in the past for electronic systems @xcite . here
the mapping is explicitly justified deep in the ordered zigzag phase [ see eq .
( [ condition ] ) ] . based on symmetry arguments
( [ tloc_gen ] ) ] we postulate that this mapping can be extended to the critical regime .
the number of universality classes in two dimensions is restricted by conformal field theory @xcite .
therefore , we argue that identifying the symmetry ( @xmath9 in our case ) determines the universality class , and implies that it is the same as of the one dimensional ising model in a transverse field . in other words , we assume that all operators obtained in our theory that differ from the ones of the conformal field theory of the one dimensional ising model in a transverse field are irrelevant in the sense of the renormalization group . to the best of our knowledge
, there are no existing theoretical tools to verify such statements .
however , this hypothesis is further supported by numerical studies @xcite . under these assumptions ,
we have estimated the quantum critical point as well as the regime of parameters where classically zigzag order would be found , but quantum tunneling and fluctuations suppress it and the linear chain is recovered . in this calculation it was assumed that the effective potential @xmath50 [ eq .
( [ v_loc ] ) ] , obtained in the vicinity of the critical point according to the classical theory , still holds .
our analysis allows us to consider possible realizations of a suitable system in which the predicted quantum phase transition would be observable .
in particular for trapped ion systems , we identify the physical parameters determining the critical point , as well as the experimental parameters which are required in order to access this regime .
schemes for measuring the transition are proposed , which are based on photon scattering .
we remark that this work addresses the two - dimensional case , in which the transition to a zigzag may only happen in a predetermined plane , and give also the full phase diagram including the thermal fluctuations . in three dimensions , where a goldstone mode is predicted corresponding to rotations of the plane , in which the zigzag is formed around the axis of the chain
, one could consider an extension of this mapping . on this basis
we conjecture that the system at the quantum phase transition can be mapped to the xy model , and the phase transition is of kosterlitz - thouless type .
we finally note that our theory provides a general framework which allows one to study further effects , such as dynamics at the quantum critical point .
it is worth mentioning that here one expects the creation of topological defects , i.e. domain walls
. it would be interesting to apply this theory to defect formation , extending the work done on the classical system in ref . , and to explore whether these systems allow for the creation of special kind of entangled states of the ions . to conclude , using a quantum - field theoretical description we argue that the linear - zigzag instability in two - dimensional systems of trapped ions or polar molecules can be mapped to the one dimensional ising model in a transverse field .
this result demonstrates once more the potentialities offered by these systems as quantum simulators @xcite , and more generally for quantum technological applications .
we gratefully acknowledge useful discussions with e. altman , e. demler , j. eschner , r. fazio , j. feinberg , m. e. fisher , y. gefen , j. meyer , m. raizen and s. sachdev . this work has been partially supported by the israel science foundation ( isf ) , by the us - israel binational science foundation ( bsf ) , by the minerva center of nonlinear physics of complex systems , by the shlomo kaplansky academic chair , by the european commission ( ip aqute , strep picc ) , by the esf ( euroquam , cmmc ) , and by the spanish ministerio de ciencia y innovacin ( qoit , consolider - ingenio 2010 ; qnlp , fis2007 - 66944 ; ramon - y - cajal ) .
e. s. acknowledges support from the ministry of science and technology ( grant 3 - 5792 ) .
g. m. acknowledges the german research council ( dfg ) for support .
in this appendix the eigenvalues of the hamiltonian @xmath199 [ eq . ( [ h0_norm ] ) ] are calculated in the semiclassical limit . in this limit
the term proportional to @xmath319 in the potential can be ignored .
the energy levels can be found by the bohr - sommerfeld rule where the action @xmath320 is quantized : @xmath321 for large @xmath113 . here ,
similarly to eq .
( [ h0_norm ] ) , we have used units where length scales are normalized by @xmath201 and energy scales by @xmath202 . for large energy ( @xmath322 ) where the semiclassical approximation is relevant , @xmath323 where @xmath324 is the conjugate momentum of the normalized coordinate @xmath325 , and @xmath326
( [ b_def ] ) ] .
the integral is over a closed trajectory of the particle . by elementary rescaling
one finds : @xmath327 note that the normalized energies scale as @xmath328 , as we found also for low energies within the approximation @xmath211 [ see eq .
( [ h_b ] ) ] .
the semiclassical energies are therefore @xmath329 the energy spacings increase with energy . in the physical units
the eigenenergies are given by @xmath330 a. del campo , g. de chiara , g. morigi , m. b. plenio , and a. retzker , phys .
lett . * 105 * , 075701 ( 2010 ) ; g. de chiara , a. del campo , g. morigi , m. b. plenio , and a. retzker , new .
* 12 * , 115003 ( 2010 ) .
cardy , _ conformal invariance and statistical mechanics , in fields , strings and critical phenomena _ , proc . of 1988
les houches summer school , e. brezin and j. zinn - justin eds .
( north - holland 1990 ) ; m. henkel , _ conformal invariance and critical phenomena _ ( springer , berlin , 1999 ) .
|
a quasi one dimensional system of trapped , repulsively interacting atoms ( e.g. , an ion chain ) exhibits a structural phase transition from a linear chain to a zigzag structure , tuned by reducing the transverse trap potential or increasing the particle density .
since it is a one dimensional transition , it takes place at zero temperature and therefore quantum fluctuations dominate . in [ fishman , _ et al .
_ , phys .
rev .
b * 77 * , 064111 ( 2008 ) ] it was shown that the system close to the linear - zigzag instability is described by a @xmath0 model .
we propose a mapping of the @xmath0 field theory to the well known ising chain in a transverse field , which exhibits a quantum critical point .
based on this mapping , we estimate the quantum critical point in terms of the system parameters .
this estimate gives the critical value of the transverse trap frequency for which the quantum phase transition occurs , and which has a finite , measurable deviation from the critical point evaluated within the classical theory .
a measurement is suggested for atomic systems which can probe the critical trap frequency at sufficiently low temperatures @xmath1 .
we focus in particular on a trapped ion system , and estimate the implied limitations on @xmath1 and on the interparticle distance .
we conclude that the experimental observation of the quantum critical behavior is in principle accessible .
|
the study of very high energy ( vhe , @xmath1 gev ) @xmath0-ray emission from active galactic nuclei ( agn ) is one of the major goals of ground - based @xmath0-ray astronomy .
the sensitivity of the current imaging air cerenkov telescopes ( iact ) enables phenomenological studies of the physics inside the relativistic jets in blazars , and in particular advances in understanding both the origin of the vhe @xmath0-rays as well as the relations between photons of different energies ( from radio to vhe ) . except for the radio galaxy m87 ,
all 17 currently known vhe gamma - ray emitting agns @xcite are blazars , rwagner / sources/ for an up - to - date source list . ] which are characterized by a close orientation of the jets with the line of sight .
their spectra are dominated by non - thermal emission that consists of two distinct broad components . while the low energy bump , located at optical to x - ray energies , is unisonously explained by synchrotron emission of electrons , the origin of the high - energy part of the spectrum is still debated .
leptonic models ascribe it to inverse compton processes that either up - scatter synchrotron photons , or to external photons that originate from the accretion disk , from nearby massive stars , or are reflected into the jet by surrounding material . in hadronic models , interactions of a highly relativistic jet outflow with ambient matter , proton - induced cascades , or synchrotron radiation off protons , are responsible for the high energy photons .
another defining property of blazars is the high variability of their emission ranging from radio to @xmath0-rays . for vhe
@xmath0-ray blazars , correlations between x - ray and @xmath0-ray emission have been found on time scales ranging from several 10 minutes to days ( e.g. , * ? ? ?
* ) , although the relation has proven to be rather complicated . here
we present selected results for magic blazar observations of mkn 421 , mkn 501 , 1es2344 + 514 , pg 1553 + 113 , and bl lacert .
magic @xcite is located on the canary island of la palma ( 2200 m a.s.l . ) and is currently the largest iact .
its energy range spans from 60 gev ( trigger threshold at small zenith angles ) up to tens of tev .
magic has a sensitivity of @xmath2 2.5% of the crab nebula flux in 50 observation hours .
its energy resolution is about 30% above 100 gev and about 25% from 200 gev onwards .
the magic standard analysis chain is described e.g. by @xcite .
observations during moderate moonshine enable a substantially extended duty cycle @xcite , which is particularly important for blazar observations .
parallel optical observations are performed with the kva 35 cm telescope .
the magic agn observation program encompasses over 500 hours / year . besides participating in various multiwavelength campaigns ( e.g. , * ? ? ? * ) and many _ target of opportunity _ observation agreements ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , magic is cooperating with other cerenkov telescopes as to extend the dynamical range and/or the time coverage of observations @xcite and monitors some well - known bright blazars on a regular basis @xcite .
apart from having observed previously known blazars , magic has also newly discovered a couple of blazars in vhe @xmath0-rays , as 1es1218 + 30.4 @xcite , mkn 180 @xcite , pg 1553 + 113 @xcite , and recently bl lacert@xcite .
mkn 421 is the closest ( @xmath3 ) known tev blazar and , along with mkn 501 , also the best studied one .
it has shown variations larger than one order of magnitude and occasional flux doubling times as short as 15 minutes @xcite .
variations in the hardness of the tev @xmath0-ray spectrum during flares were reported by several groups ( e.g. , * ? ? ?
* ; * ? ? ?
simultaneous observations in the x - ray and vhe bands showed a significant correlation of the respective fluxes @xcite .
mkn 421 was observed with magic during 19 nights for a total of 25.6 hours , with observation times per night ranging from 30 minutes up to 4 hours .
it was found to be in a medium flux state ranging from 0.5 to 2 crab units above 200 gev .
significant variations of up to an overall factor of four and up to a factor two between successive nights were found . a clear correlation between the @xmath4 kev x - ray count rate , obtained from the public rxte all - sky monitor data , and the vhe @xmath0-ray flux can be described well by both a linear fit ( correlation coefficient @xmath5 ) and a parabolic fit ( both forced to go through the origin ) .
the energy spectrum of mkn 421 , for the first time measured down to @xmath6 gev , is shown in fig .
[ fig:421:3 ] along with the reconstructed intrinsic spectrum .
the latter has been obtained by correcting for @xmath7 absorption effects caused by extragalactic background light ( ebl ; hauser & dwek 2001 ) , using the recent ebl model of @xcite .
the intrinsic spectrum is clearly curved , implying that the curvature in the measured spectrum has an intrinsic origin rather than being caused by ebl absorption .
both a power - law+exponential cutoff fit and a curved power - law fit indicate a flattening of the spectrum towards 100 gev .
the spectrum is hardening with an increasing flux level ( fig .
[ fig:421:4 ] ) .
the analysis is described in detail in @xcite .
mkn 501 ( @xmath8 ) is known to be a strong and variable vhe @xmath0-ray emitter . during a flare in 1997
mkn 501 showed strong variability on timescales of 0.5 days , and the integral flux reached 10 times the flux of the crab nebula above @xmath9 tev ( e.g. , * ? ? ? * ) .
while a correlation between spectral hardness and @xmath0-ray flux was found , the position of the inverse compton peak could not be detected .
magic observed mkn 501 for 24 nights during six weeks in summer 2005 .
about 18 hours out of the total 30 observation hours were performed in the presence of ( moderate ) moonshine . during most of the observations
the source was in a rather low flux state , corresponding to 30% of the crab nebula flux at @xmath10 gev . in two nights ( one with moon present ) , however , mkn 501 was reaching up to four times the crab nebula flux .
a correlation between spectral hardness and vhe @xmath0-ray flux was found also for mkn 501 ( fig .
[ fig:501:4 ] ) . for the flare nights , a clear peak in the spectrum
is seen ( e.g. , fig . [ fig:501:5 ] ) . in leptonic models
it can be identified with the `` inverse compton peak '' , which now for the first time is seen unambiguously in vhe range for blazars
. a closer look at the two flare nights reveals rapid flux changes with doubling times as short as 3 minutes or less ( fig .
[ fig:501:2 ] ) , while constant background rates assure that no instrumental effects are responsible for these .
[ fig:501:3 ] shows corresponding light curves in different , distinct energy bands between 150 gev and 10 tev .
interestingly the flares in the two nights behave differently : while the 2005 june 30 flare is only visible in 250 gev to 1.2 tev , the 2005 july 09 flare is apparent in all energy bands , and a marginal time delay towards higher energies is visible . for the first time , short ( @xmath11 20 min ) flares with a resolved time structure can now be subjected to detailed studies of particle acceleration and cooling timescales .
the analysis and more results are shown in @xcite .
1es2344 + 514 ( @xmath12 ) was detected in vhe @xmath0-rays in 1995/6 on the @xmath13 level @xcite . over 1/3 of the excess
was recorded during one night in december 1995 with @xmath14 64% of the crab nebula flux .
the @xmath0-ray emission was confirmed on a @xmath15 significance level by @xcite in observations in 1997 - 2002 . in 2005 , magic saw a clear @xmath16 excess from the direction of 1es2344 + 514 .
the measured flux corresponds to 10% of the crab nebula flux ( @xmath17 gev ) . for the first time a diurnal measurement of the emission of this blazar was possible , yielding no strong evidence for significant variability on times scales of days and thus giving first insights to the day - scale properties of a low blazar emission state .
the differential energy spectrum can be fitted by a simple power law with a photon index of @xmath18 .
the derived spectrum is softer than the one reported by the whipple collaboration during the flare in 1995 @xcite .
details on the analysis are given in @xcite .
based on its sed properties , pg 1553 + 113 was one of the most promising candidates for vhe @xmath0-ray emission @xcite .
the blazar was observed in 2005/6 , yielding a @xmath0-ray signal at a significance level of @xmath19 .
there is no evidence for short term variability on a time scale of days , but a significant change by a factor of three in the flux level from 2005 to 2006 was found . the vhe @xmath0-ray flux was not found to correlate with a substantial brightening of the source in the optical @xmath20-band around mjd 53820 ( fig . [ fig:1553:2 ] ) .
the differential energy spectrum for pg 1553 + 113 is well described by a pure power law with a photon index of @xmath21 , the steepest photon index observed so far for blazars ( fig .
[ fig:1553:3 ] ) .
the signal detected by magic confirms a tentative signal seen by h.e.s.s . at a higher energy threshold with data taken at about the same time as magic in 2005 @xcite .
the source , therefore , can now be considered as a firmly established vhe @xmath0-ray emitter .
details of the analysis are given in @xcite .
the distance of pg 1553 + 113 is currently unknown , despite several attempts of determining its redshift ( direct measurements : @xmath22 , * ? ? ?
the objects shows no emission lines , which probably implies a close alignment of the jet with the line of sight , or a jet outshining the galaxy core . using the observed spectrum of pg 1553 + 113 and assuming a minimum possible density of the evolving ebl ( lower limit in * ? ? ?
* ) , an upper limit of @xmath23 can be inferred when requiring the intrinsic photon index @xmath24 .
other works employ sed fits @xcite , the structure of the intrinsic spectrum @xcite , or blazar ensemble arguments @xcite to determine redshift limits .
bllacert ( @xmath25 ) is the historical prototype of the bl lac class of agns . unlike for the up to recently known `` vhe blazars '' with their synchrotron peaks in the x - ray domain ,
the peak of its synchrotron emission is located in the sub - millimeter to optical band ( @xmath26 hz ) .
22.2 hours of magic observations of bl lacert from august to december 2005 resulted in a @xmath27 vhe @xmath0-ray signal . above 200 gev ,
an integral flux of @xmath28 was measured , corresponding to approximately 3% of the crab nebula flux .
no significant evidence of flux variability was found .
optical and radio data shows enhanced flux levels during the observations ; while the vhe @xmath0-ray flux shows similar variations , no correlation can be claimed at this point .
the differential spectrum between 150 and 900 gev was found to be rather steep , with a photon index of @xmath29 .
in contrast to a 1997 flare of bllacert , which necessitated external photon fields to explain the vhe @xmath0 emission @xcite , the magic observations can be accommodated in a simple ssc model .
the detection is described in @xcite .
ackermann , m. , et al .
( icecube & magic collabs . ) , 2007 , in proc .
30th icrc ( mrida ) aharonian , f. , et al .
( hegra collab . ) , 1999 , a&a , 342 , 69 aharonian , f. , et al .
( hegra collab . ) .
2004 , a&a , 421 , 529 aharonian , f. , et al .
( h.e.s.s .
collab . ) , 2005 , a&a , 442 , 895 aharonian , f. , et al .
( h.e.s.s .
collab . ) , 2006 , a&a , 448 , l19 albert , j. , et al .
( magic collab . ) , 2006a , apjl , 642 , l119 albert , j. , et al .
( magic collab . ) , 2006b , apjl , 648 , l105 albert , j. , et al .
( magic collab . ) , 2007a , apjl , 654 , l119 albert , j. , et al .
( magic collab . ) , 2007b , apj , 662 , 892 albert , j. , et al .
( magic collab . ) , 2007c , apj , 663 , 125 albert , j. , et al .
( magic collab . ) , 2007d , apj accepted , arxiv : astro - ph/0702008 albert , j. , et al .
( magic collab . ) , 2007e , aph subm .
, arxiv : astro - ph/0702475 albert , j. , et al .
( magic collab . ) , 2007f , apjl subm .
, arxiv : astro - ph/0703048 badran , h. m. & weekes , t. c. , 2001 , in proc .
27th icrc , hamburg , 2653 baixeras , c. , et al .
( magic collab . ) , 2004 , nucl .
instrum . meth . , a518 , 188 catanese , m. , et al .
( whipple collab . ) , 1998 , apj , 501 , 616 cortina , j. et al .
( magic collab . ) , 2005 , in proc .
29th icrc ( pune ) , 5 , 359 costamante , l. , et al . , 2001 , a&a , 371 , 512 costamante , l. & ghisellini , g. , 2002 , a&a , 384 , 56 fossati , g. , et al . , 2004 , newar , 48 , 419 gaidos , j. a. , et al .
( whipple collab . ) , 1996 , nature , 383 , 319 goebel , f. , et al .
( magic collab . ) , 2007 , in proc .
30th icrc ( mrida ) hauser , m. g. @xmath30 dwek , e. , 2001 , ara&a , 39 , 249 hayashida , m. , et al .
( magic collab . ) , 2007 , in proc 30th icrc ( mrida ) kneiske , t. m. , et al . , 2004 , a&a , 413 , 807 krawczynski , h. , et al . , 2001 ,
apj , 559 , 187 krennrich , f. , et al . , 2002 ,
apj , 575 , l9 mazin , d. & goebel , f. , 2007 , apjl , 655 , l13 mazin , d. , et al .
( h.e.s.s .
& magic coll . ) , 2005 , in proc .
29th icrc ( pune ) , 4 , 331 mazin , d. , et al .
( magic collab . ) , 2007 , in proc 30th icrc ( mrida ) primack , j. , et al . , 2005 , aip conf .
, 745 , 23 ravasio , m. , et al . , 2002 , a&a , 383 , 763 sbarufatti , b. , et al .
, 2006 , aj , 132 , 1 schroedter , m. , et al . , 2005 , apj , 634 , 947 wagner , r. m. , 2007a , these proceedings wagner , r. m. , 2007b , 2007 , in proc .
30th icrc ( mrida ) , arxiv:0706.4442 [ astro - ph ] wagner , r. m. , et al .
( magic collab . ) , 2005 , in proc .
29th icrc ( pune ) , 4 , 163
|
magic currently is the largest imaging atmospheric cerenkov telescope world - wide . since 2004 , @xmath0-ray emission from several active galactic nuclei in the gev / tev energy range has been detected , some of which were newly discovered as very - high energy @xmath0-ray sources .
the @xmath0-rays are assumed to originate from particle acceleration processes in the agn jets .
we give an overview of the agn observed and detected by magic , discuss spectral and temporal properties of these and show physics implications of some selected observations .
|
the gravastar model has been proposed by mazur and mottola ( mm ) as a possible end point of the gravitational collapse of a massive body @xcite .
as opposed to the classical scenario of the gravitational collapse into a schwarzschild black hole , the gravastar concept avoids the formation of the black hole horizon . technically , the gravastar is a global , static , spherically symmetric solution to the einstein equations . in the interior
there is a segment of the de sitter geometry which is matched to the exterior schwarzschild geometry by means of a spherical shell of matter . in the original mm model the quantum phase transition takes place within the boundary layer / shell , but the full quantum treatment is immediately replaced by the mean field approximation in which ( classical ) einstein equations are valid .
the de sitter geometry in the interior of the gravastar implies constant positive ( dark ) energy density @xmath2 accompanied by the isotropic negative pressure @xmath3 .
such energy - momentum content of the spacetime is equivalent to introducing the positive cosmological constant @xmath4 into the einstein equations , and can be understood as the gravitational vacuum , or gravitational condensate , which motivated the coining of the term gravastar ( _ _ gra__vitational _ _ va__cuum _ star _ ) . in this original model of mm ,
the phase transition layer consists of two spherical @xmath0-shells with vanishing surface energy density but with non - vanishing surface pressure , between which there is a thin layer of stiff matter ( @xmath5 ) .
the transition layer is positioned arbitrarily close to where the horizon of the schwarzschild geometry is expected to form . in this way
the gravastar model allows for arbitrarily high gravitational surface red - shifts and is , for an external observer , indistinguishable from being a true black hole .
a simplified gravastar model was constructed by visser and wiltshire @xcite in order to study the stability properties .
the phase transition layer was replaced by a single spherical @xmath0-shell .
the issue of stability could not be fully settled because the equation of state for the shell is not available .
however , it was shown that physically reasonable equations of state could lead to gravastar configurations that are stable against radial perturbations .
this simplified gravastar model also motivated a number of authors to consider different vacuum geometries in the interior and the exterior .
this includes the models with anti - de sitter interior and the schwarzschild
de sitter exterior geometries and also the reissner
nordstrm geometries @xcite , the born infeld phantom gravastar @xcite , and the non - linear electrodynamics gravastar @xcite . a different development of the gravastar idea went in the direction of constructing models with continuous profiles for the energy density and the anisotropic pressures @xcite . in these models
the energy momentum tensor in the centre of the spherical structure satisfies @xmath6 and through the layer of anisotropic principal pressures ( crust ) smoothly joins with the exterior vacuum spacetime .
gravastars with shells of finite thickness but without @xmath0-shells at the boundaries with the vacuum regions have been considered in ref .
@xcite and it has been pointed out that quasi - normal modes of oscillations might be a possible way of distinguishing gravastars from black holes .
an account of the observational constraints placed on the gravastar model has been given in ref .
the gravastar models are also closely related to a broader subject of dark energy stars , see eg
. ref .
in fact , the ideas of astrophysical bodies with the de sitter core representing the gravitationally collapsed matter at extremely high density can be traced back to the work of sakharov @xcite and gliner @xcite and have been further investigated by dymnikova in a series of papers starting with @xcite , see eg .
refs @xcite and references therein .
given that the physical mechanism behind the proposed vacuum phase transition within the gravastar shell is essentially unknown , the energy conditions of gr @xcite provide the most widely accepted framework of model - of - matter independent constraints one can impose on the ` exotic matter ' comprising the shell .
it is therefore important to fully investigate the constraints on the gravastar model that are imposed by the energy conditions . in this work
we will consider the @xmath0-shell gravastar model of ref @xcite and we will derive the sharp analytic bounds on the surface compactness that follow from the requirement that the dominant energy condition ( dec ) is satisfied at the shell .
the dec will include the weak ( wec ) and the null energy conditions ( nec ) , but not the strong energy condition ( sec ) .
since the sec is violated from the outset by the dark energy in the interior of the gravastar , it does not appear natural to require that it holds throughout the phase transition layer . in sec .
[ sec : model ] we review the @xmath0-shell gravastar model and we introduce the configuration variables that we use through the rest of the paper . in sec .
[ sec : ec ] we discuss the energy conditions of gr imposed on the energy - momentum tensor of the gravastar shell and in sec .
[ sec : allowed ] we carry out the detailed analysis of the gravastar configurations allowed by the dec . in sec .
[ sec : eos ] we discuss the properties of the equation of state as derived from the assumed stability of the gravastar with respect to radial perturbations .
conclusions are given in sec .
[ sec : conc ] .
the line element for the interior and the exterior geometries can be written using the geometrized units @xmath7 and the coordinates @xmath8 in the form @xmath9 where @xmath10 is the metric on the unit 2-sphere , @xmath11 is the time coordinate scaling constant , and @xmath12 is the _ compactness function _ related to the quasi - local mass function @xmath13 by @xmath14 where @xmath15 is the energy density .
in general , to avoid the formation of the event horizon within the spherical body the compactness must be less than unity .
in the interior of the gravastar shell we have the segment of the de sitter geometry with the constant energy density @xmath16 and the compactness is given by @xmath17 at @xmath18 the interior compactness reaches unity which corresponds to the position of the de sitter ( cosmological ) horizon .
therefore the surface of the gravastar must be located at the radius @xmath19 . in the exterior
we have the schwarzschild de sitter geometry with the constant energy density @xmath20 and mass parameter @xmath21 .
the exterior compactness is given by @xmath22 with @xmath23 , @xmath21 , this is clearly the schwarzschild geometry , while with @xmath24 , @xmath25 , this is the de sitter geometry . in the general case we are considering , @xmath24 , @xmath21
, the gravastar surface must be located in the region where @xmath26 . for this region to exist the condition @xmath27 must hold , whereupon the radial coordinates of the two horizons of the schwarzschild de sitter geometry which bound the region are given by the roots of @xmath28 ( a cubic ) .
thus , if the gravastar surface is located at @xmath29 such that @xmath30 and @xmath31 , no horizon forms within the gravastar geometry up to the outer horizon of the schwarzschild de sitter geometry . since the original motivation behind the mm gravastar model was to construct a highly compact object that could represent a gravitationally collapsed body of positive gravitational mass , one expects the ( dark ) energy density inside the gravastar to be greater than that of the surrounding space , i.e. @xmath32 , or equivalently @xmath33 .
we refer to the above as the `` gravastar requirement '' .
note also that the constants @xmath34 and @xmath35 are introduced for notational convenience ; however , @xmath35 can be viewed as the cosmological constant @xmath36 present in the einstein equations provided that in the exterior of the gravastar shell we set @xmath37 and in the interior we set @xmath38 .
thus , in the picture with the cosmological constant , demanding that the energy density in the interior of the gravastar shell is non - negative , @xmath39 , is equivalent to our `` gravastar requirement '' . therefore
, although it _ is _ technically possible to construct the gravastar - like solutions to the einstein equations with @xmath40 and @xmath21 , we will not consider them in this analysis . through
the rest of the paper we will describe the gravastar configurations in terms of the configuration variables defined as follows : @xmath41 @xmath42 and @xmath43 are the values of the compactness on the interior and the exterior side of the shell , while @xmath44 is the contribution to the compactness at the exterior side of the shell due to the energy density in the exterior . for the gravastar geometry we require @xmath45 @xmath44 and @xmath42
are non - negative since the energy densities in the interior and in the exterior are non - negative by assumption .
the condition @xmath46 reflects the above mentioned ` gravastar requirement ' .
the condition @xmath47 ensures that the mass of the gravastar , @xmath48 , is non - negative , while @xmath49 , as will be shown in sec .
[ sec : ec ] , ensures that the energy density of the gravastar shell is non - negative .
the gravastar shell is a timelike hypersurface of constant radius @xmath29 along which the interior and the exterior geometries are joined .
the israel s thin shell formalism @xcite ( for textbook coverage see eg .
@xcite ) allows one to assign the energy - momentum distribution to the hypersurface , as required to make the resulting spacetime a solution of einstein equations .
the first junction condition requires that the induced metric on the hypersurface be the same for the metrics on both sides of the hypersurface . parametrizing the hypersurface with the coordinates @xmath50 , the hypersurface metric tensor @xmath51 as induced from the metric ( [ eq : gends2 ] ) is given by @xmath52 , where @xmath53 .
the hypersurface line element is given by @xmath54 therefore , for the first junction condition to hold we must have @xmath55 .
this is accomplished by setting @xmath56 for compatibility with the usual form of the exterior schwarzschild metric in case @xmath23 , and setting @xmath57 to appropriately rescale the time coordinate in the interior .
the second junction condition requires that for the smooth joining of the metrics on the two sides of the hypersurface the extrinsic curvature on the hypersurface , @xmath58 , be the same on both sides of the hypersurface .
the nonvanishing components of the extrinsic curvature for the hypersurface of radius @xmath29 embedded in the metric of the form ( [ eq : gends2 ] ) are @xmath59 in the gravastar the second junction condition is not satisfied .
the joining of the metrics is not smooth , but is still possible if one allows for the energy - momentum distribution on the hypersurface .
the standard expression for the shell energy - momentum tensor on a timelike hypersurface is @xmath60 - \big [ k \big ] \delta^a_b \right),\ ] ] where the brackets denote the discontinuity of the quantity over the hypersurface and @xmath61 . using the compactness functions for the exterior and interior metric given by ( [ eq : muint ] ) and ( [ eq : muext ] ) and the notation ( [ eq : xyz ] ) ,
the surface energy density can be written @xmath62 = \frac{1}{4\pi\,a } \ , \left ( \sqrt{1-x } - \sqrt{1-y } \right ) , \label{eq : sigma}\ ] ] and the isotropic surface tension can be written @xmath63 \nonumber \\ & = \frac{1}{8\pi\,a } \ , \left ( \frac{1 - 2x}{\sqrt{1-x } } - \frac{1-\frac12y-\frac32z}{\sqrt{1-y } } \right ) .
\label{eq : theta } \end{aligned}\ ] ] note that by definition the surface tension has the opposite sign of the surface pressure , i.e. , @xmath64 indicates that the surface ( shell ) is ` under tension ' , while @xmath65 indicates that it is ` under pressure ' .
the four standard energy conditions of gr formulated in terms of the components of the energy - momentum tensor of the gravastar shell the surface energy density @xmath66 ( [ eq : sigma ] ) and isotropic surface tension @xmath67 ( [ eq : theta ] ) read as follows * weak energy condition ( wec ) : @xmath68 and @xmath69 , * null energy condition ( nec ) : @xmath69 , * strong energy condition ( sec ) : @xmath69 and @xmath70 , * dominant energy condition ( dec ) : @xmath68 and @xmath71 .
one can see from the structure of ( [ eq : sigma ] ) and ( [ eq : theta ] ) that the status of the energy conditions for the gravastar shell can be studied in the space of the three configuration variables , @xmath72 .
as pointed out in introduction , we require that the energy momentum tensor of the gravastar shell satisfies all energy conditions except the sec .
since the dec is more restrictive than the wec or the nec , we proceed to consider the dec through a number of steps .
we begin with the condition of non - negative surface energy density which , as can be easily seen from ( [ eq : sigma ] ) , reduces to @xmath73 this requirement has already been encoded in ( [ eq : zxy ] ) .
continuing with the condition @xmath71 , we break it into cases according to @xmath74 .
we find @xmath75 where the function @xmath76 is given by @xmath77 the configurations with @xmath78 (
vanishing surface tension / pressure ) , satisfying @xmath79 , are understood as the shells of dust .
the dec is clearly satisfied for dust shells provided ( [ eq : nnsigma ] ) holds . for the case @xmath80 ,
i.e. shells under ( positive ) surface pressure , the condition @xmath71 reduces to @xmath81 where @xmath82 is given by @xmath83 for the case @xmath64 , i.e. shell under ( positive ) tension , @xmath71 reduces to @xmath84 where @xmath85 combining the above cases with the condition ( [ eq : nnsigma ] ) allows us to summarize the dec ( which includes the wec and the nec ) with the following conditions on the configuration variables @xmath72 : @xmath86 the configurations saturating the dec with @xmath87 and @xmath88 can be called stiff and anti - stiff shells .
the implications of these conditions on the allowed gravastar configurations are studied in the next section .
the shape of the functions @xmath76 , @xmath82 and @xmath89 , given by ( [ eq : fzero ] ) , ( [ eq : fplus ] ) and ( [ eq : fminus ] ) , determines the regions in the space of configuration variables @xmath72 that are allowed by the dec ( [ eq : dec ] ) .
in addition , we recall the ` gravastar condition ' @xmath46 , expressing the requirement that the ( dark ) energy density inside the gravastar is not less than the energy density in the exterior spacetime . before dividing the analysis into cases corresponding to specific ranges of the configuration variable @xmath44
, we observe some general properties of @xmath76 , @xmath82 and @xmath89 . as already pointed out in sec .
[ sec : model ] , @xmath90 implies @xmath91 ( see eq .
( [ eq : muext ] ) ) , but it can also be shown that @xmath92 i.e. , the dec ( [ eq : dec ] ) can not be satisfied with @xmath90 . for @xmath93 , which corresponds to @xmath25
, one can show that @xmath94 indicating that the only @xmath93 configuration satisfying the dec is @xmath95 , which further implies @xmath96 , i.e. , simply de sitter spacetime with no gravastar .
for @xmath97 it can be shown that the following relations hold , @xmath98 this indicates that the dec satisfying configurations with @xmath65 ( shell under surface pressure ) exist for all @xmath44 . in fig .
[ fig : one ] the dec satisfying gravastar configurations are indicated with the shaded regions in the @xmath99 plane for several fixed values of @xmath44 .
details are discussed in what follows . with @xmath100
we are actually considering the gravastar with the schwarzschild exterior geometry .
one can show that @xmath101 for @xmath102 , meaning that in this case the dust shells and the shells under surface tension can not satisfy the condition ( [ eq : zxy ] ) .
inspection of @xmath103 reveals that it is greater than zero for @xmath104 with a maximum at @xmath105 , @xmath106 it follows that the @xmath100 gravastar configurations allowed by the dec involve the shell under surface pressure and are constrained by @xmath107 the corresponding region in the @xmath99 plane is shown in fig .
[ fig : one ] ( upper - left plot ) . in the original mazur
mottola gravastar model it is assumed that the compactness on both sides of the vacuum phase transition layer will approach unity ( horizon formation ) to induce the vacuum phase transition , i.e. @xmath108 . as we can see from the above analysis , the dec satisfying configurations are clearly pushed away from the mazur
mottola limit .
the highest compactness on the exterior side of the gravastar shell that can be reached without violating the dec is @xmath109 .
it corresponds to the surface gravitational redshift @xmath110 , and is obtained with flat interior geometry . the same upper bound on the surface compactness was obtained in a similar context @xcite where the dec - satisfying shell was considered outside of a schwarzschild black hole of mass @xmath111 , in the limit where @xmath112 .
the finding which is relevant in the context of the gravastar model is that the introduction of the de sitter geometry ( dark energy ) in the interior , relative to the flat interior geometry ( hollow shell ) , does not help to support the dec - satisfying shell against the collapse at high values of the surface compactness , as it might be naively expected due to the repulsive character of the de sitter interior ( free falling particle accelerating away from the centre ) .
we have also obtained the upper bound on the compactness at the interior side of the shell ; the highest @xmath42 that can be reached without violating the dec is @xmath113 , given by ( [ eq : z0xmax ] ) , lowering the maximal exterior compactness to @xmath105 ( surface redshift @xmath114 ) .
let us also note that in the model of mazur and mottola the vacuum phase transition shell is contributing a negligible fraction to the net gravitational mass of the gravastar .
here we see that such a condition can not be met if the dec is to be satisfied . with @xmath116
we are introducing the schwarzschild de sitter exterior geometry .
let us begin with the configurations with the gravastar shell under ( positive ) surface pressure ( @xmath65 ) .
the functions @xmath76 and @xmath82 tend to @xmath117 as @xmath118 , but they have a maximum in @xmath43 with the value greater than @xmath44 for @xmath119 .
the area in the @xmath99 plane bounded by @xmath76 and @xmath82 is the @xmath65 dec satisfying region , but we must also observe the ` gravastar requirement ' @xmath120 .
therefore we look for the intersections of @xmath76 and @xmath82 with the @xmath121 line and we find @xmath122 and @xmath123 the dec satisfying gravastar configurations with the shell under surface pressure ( @xmath65 ) are therefore @xmath124 where @xmath125 and @xmath126 are given by ( [ eq:2y0 ] ) and ( [ eq:2y+ ] ) .
the dust shells ( @xmath78 ) configurations are those with @xmath127 upon showing that @xmath128 for @xmath129 it follows that the dec satisfying configurations with the shell under surface tension ( @xmath64 ) are @xmath130 the dec satisfying gravastar configurations for the case @xmath131 are shown in fig .
[ fig : one ] ( upper - right plot ) .
considering @xmath116 puts the gravastar model into the context where positive cosmological constant @xmath132 is present in the einstein equations . here
we can no longer discuss the surface gravitational redshift , but we still find it important to discuss the maximal surface compactness that can be reached without violating the dec . as in the @xmath100 case ,
the highest compactness on the exterior side of the shell is achieved with the ` hollow interior ' , @xmath121 , which here means that the ( dark ) energy density in the interior is equal to that in the exterior ; the maximal compactness is @xmath133 given in ( [ eq:2y+ ] ) .
it can be shown that @xmath133 starts with the value @xmath134 at @xmath100 and increases toward unity as @xmath135 . however
, if the ( dark ) energy density in the interior is higher than that in the exterior region , @xmath136 , the maximal compactness that can be achieved is lower than that given by @xmath126 . another interesting feature of gravastar configurations with @xmath116 is that in addition to the configurations with shell under surface pressure here we can have shells of dust and shells under surface tension .
the @xmath137 case can be seen as the critical case between the regimes @xmath138 .
the functions @xmath139 and @xmath140 , bounding the @xmath65 dec satisfying region , are monotonically increasing in @xmath43 for @xmath141 , reaching finite values @xmath142 and @xmath143 , while @xmath144 is monotonically decreasing diverging to @xmath117 as @xmath118 . observing also the ` gravastar requirement ' @xmath145 ,
the dec satisfying configurations are simply @xmath146 and are shown in fig .
[ fig : one ] ( lower - left plot ) .
the distinctive feature of this special case @xmath137 is that there is a wide range of configurations that allow the surface compactness on the exterior side to approach unity arbitrarily close .
the compactness on the interior side of the shell is still bounded from above by @xmath82 , or fixed by @xmath76 for a dust shell .
for example , the dust shell with @xmath118 will have @xmath147 , which implies interior ( dark ) energy density which higher by factor @xmath148 than the ( dark ) energy density in the exterior .
however , the value of @xmath137 considered here implies that the radius of the gravastar is larger than one half of the radius of the exterior horizon of the schwarzschild
de sitter geometry , which makes these configurations highly unlikely to be relevant in the astrophysical context . for completeness here
we consider also the @xmath149 configurations .
the functions @xmath89 , @xmath76 and @xmath82 tend to @xmath150 as @xmath151 .
since @xmath152 for @xmath141 , the configurations with dust shells and shells under surface pressure ( @xmath153 ) are @xmath154 the shell under tension ( @xmath64 ) configurations must be treated more carefully since it can be shown that the function @xmath155 dips into the @xmath156 region for @xmath157 , which is forbidden by the ` gravastar requirement ' @xmath46 .
we find @xmath158 so for @xmath157 , the @xmath64 dec satisfying gravastar configurations are @xmath159 where @xmath160 is given by ( [ eq:4y- ] ) .
for @xmath161 the @xmath64 dec satisfying gravastar configurations are @xmath162 the special case @xmath163 is shown in fig .
[ fig : one ] ( lower - right plot ) .
the interesting feature of @xmath164 configurations is that the compactness on both sides of the shell can approach unity without violating the dec .
however , as already commented in the case of @xmath137 configurations , the size of such object is too large to represent a body of astrophysical interest .
gravastar configurations allowed by the dec shown as shaded regions in the @xmath99 plane for fixed values of @xmath165 .
( @xmath166 are defined in ( [ eq : xyz])),title="fig : " ] gravastar configurations allowed by the dec shown as shaded regions in the @xmath99 plane for fixed values of @xmath165 .
( @xmath166 are defined in ( [ eq : xyz])),title="fig : " ] + gravastar configurations allowed by the dec shown as shaded regions in the @xmath99 plane for fixed values of @xmath165 .
( @xmath166 are defined in ( [ eq : xyz])),title="fig : " ] gravastar configurations allowed by the dec shown as shaded regions in the @xmath99 plane for fixed values of @xmath165 .
( @xmath166 are defined in ( [ eq : xyz])),title="fig : " ]
in a perfect fluid described by an equation of state of the form @xmath167 , where @xmath168 is the isotropic pressure and @xmath2 is the energy density , the speed of sound propagation , @xmath169 , is given by the relation @xmath170 . in order for the causality to be preserved , it is natural to require that the sound speed does not exceed the speed of light , i.e. in our units @xmath171 .
for a view on the status of the relation among causality and the speed of sound see ref .
@xcite . here
we will investigate the speed of sound propagating ( tangentially ) along the gravastar shell , assuming that the equation of state for matter comprising the shell can be given in the simple ( barotropic ) form @xmath172 .
note that while the matter comprising the shell can be considered as the perfect fluid on the hypersurface , in the complete spacetime it represents a highly anisotropic structure ( the transverse pressure is described by the @xmath0-distribution ) . in ref .
@xcite a procedure has been developed that allows extracting of the equation of state @xmath172 of the matter comprising the gravastar shell on the basis of the requirement that the shell be stable against radial perturbations . from the dynamical version of the expressions for the surface energy density and the surface tension of the shell , which involve the terms @xmath173 and @xmath174 , overdot indicating the derivative with respect to the proper time of the observer co - moving with the shell , the ` classical ' equation of motion of the form @xmath175 follows .
the ` potential ' @xmath176 is a function of the gravastar radius @xmath177 and other configuration variables , while the ` energy ' @xmath178 ( one is not allowed to adjust the energy as in the classical context ) . upon identifying @xmath179 and @xmath180
( see ref .
@xcite for the full derivation ) the dynamical expressions for @xmath66 and @xmath67 become functions of the dynamical gravastar radius .
the requirement for the stability of a shell at the radius @xmath177 can then be expressed as the requirement imposed on the potential , @xmath181 therefore , if one chooses a particular shape of the potential @xmath182 leading to stable gravastar shells , the equation of state @xmath172 can be parametrically extracted by varying @xmath177 over the range of interest .
the important special case is @xmath183 which , as argued in ref .
@xcite , reflects the notion of strict stability of the shell .
if the shell would be displaced to a nearby radius , it would find itself there in a new state of equilibrium . in the case
@xmath183 the dynamical equations for surface energy density and surface pressure of the shell coincide with ( reduce back to ) the static expressions for @xmath66 and @xmath67 given by ( [ eq : sigma ] ) and ( [ eq : theta ] ) .
therefore we can use this approach to compute the quantity @xmath184 directly from ( [ eq : sigma ] ) and ( [ eq : theta ] ) .
we now proceed to compute @xmath185 for the gravastar models we have considered so far . using ( [ eq : cs2def ] ) , ( [ eq : sigma ] ) and ( [ eq : theta ] ) one can obtain an expression for @xmath185 in terms of @xmath177 , @xmath34 , @xmath35 and @xmath48 which is rather involved , but with the use of our configuration variables @xmath166 defined in ( [ eq : xyz ] ) it becomes much more compact and can be written @xmath186 } { 4 u^2 v^2 [ - 2 v + u ( - 1 + 3 v^2 + 3 z ) ] } \label{eq : cs2uvz}\ ] ] where @xmath187 and @xmath188 . the quantity @xmath185 diverges for @xmath189 and @xmath190 and also at @xmath191 . to obtain the boundaries of the regions in the @xmath99 plane ( for the given values of @xmath44 )
in which @xmath192 one must solve @xmath193 for @xmath194 which leads to a cubic equation .
the resulting expressions are very complicated so we only plot the results for @xmath100 ( schwarzschild exterior geometry ) and @xmath137 ( case considered in sec .
[ sec:43 ] ) in fig .
[ fig : two ] .
the region where causality is preserved with the equation of state having usual properties ( surface pressure increases as energy density increases ) are indicated by ` a ' .
we can see that the causality requirement combined with the energy conditions considered before imposes still stronger upper bounds on the gravastar exterior surface compactness @xmath43 . as an example , for the @xmath195 case ( hollow shell , schwarzschild exterior ) the highest allowed surface compactness is @xmath196 or analytically @xmath197 where @xmath198 . in both examples shown in fig .
[ fig : two ] we also see that the region ` c ' where @xmath185 is negative includes some regions that are allowed by the energy conditions .
negative @xmath185 means that the surface pressure decreases as the surface energy density increases , which indicates instability with respect to perturbations .
therefore , the requirement that the dependence of @xmath67 on @xmath66 is such that stability with respect to radial perturbations is obtained can lead to configurations where the gravastar shell is unstable against tangential perturbations such as sound propagation .
behaviour of the quantity @xmath199 ( speed of sound squared ) as derived from the requirement for the stability against the radial perturbations for the gravastar shells with @xmath100 ( left plot ) and @xmath137 ( right plot ) , shown in the @xmath99 plane . dashed lines indicate configurations where @xmath200 , while the solid lines bound the region ` a ' in which @xmath201 .
region ` b ' is the @xmath202 superluminal regime , while ` c ' is the @xmath203 unstable regime .
shaded regions are the gravastar configurations allowed by the dec ( see fig . [
fig : one]).,title="fig : " ] behaviour of the quantity @xmath199 ( speed of sound squared ) as derived from the requirement for the stability against the radial perturbations for the gravastar shells with @xmath100 ( left plot ) and @xmath137 ( right plot ) , shown in the @xmath99 plane .
dashed lines indicate configurations where @xmath200 , while the solid lines bound the region ` a ' in which @xmath201 . region ` b ' is the @xmath202 superluminal regime , while ` c ' is the @xmath203 unstable regime
shaded regions are the gravastar configurations allowed by the dec ( see fig .
[ fig : one]).,title="fig : " ]
compact objects like gravastars envisaged as possible alternatives to black holes are certainly not less controversial than black holes themselves @xcite .
the essential phenomenological feature of a gravastar that makes it difficult to distinguish such a body from a true black hole is its high surface redshift . also , the analysis of radiation from a supermassive object sgr a@xmath204 @xcite implied the existence of a horizon in this object . however , the absence of detectable heating could also be consistent with a gravastar possesing sufficiently large heat capacity @xcite .
another feature which may help in this respect is the analysis of quasi - normal modes of oscillation of a gravastar .
they are found to be different from those of a black hole @xcite , and could , in principle , provide a way to observationally discern a gravastar from a black hole . in the same spirit
it is important to analyze the constraints on possible realisations of gravastar models .
since the physical theory behind the quantum vacuum phase transitions which are assumed to take place within the gravastar shell has not been fully formulated , the energy conditions of gr provide the most general model - independent constraint on the structure of the energy - momentum tensor of the gravastar .
we have carried out a detailed analysis of the constraints that the energy conditions impose on the possible configurations of the single @xmath0-shell version of the gravastar model @xcite .
the most relevant constraint is the upper bound on the surface compactness @xmath205 of the gravastar which was in the original mm model assumed to approach unity arbitrarily close .
exactly this feature gives raise to arbitrarily large surface gravitational redshift which makes the gravastar observationally difficult to distinguish from the true schwarzschild black hole .
( note that the original mm model violates the dec at its two @xmath0-shells that are assumed to possess nonvanishing surface pressure but vanishing surface energy density . for a general discussion of known violations of energy conditions
refs @xcite ) .
in this paper we have derived the bounds on the surface compactness that follow from the requirement that the matter comprising the gravastar shell satisfies the dec .
the procedure is fully analytic and has been made possible by the use of the configuration variables @xmath166 based the values of the compactness at the inner and the outer side of the shell defined in ( [ eq : xyz ] ) .
the space of gravastar configurations satisfying the dec is confined within the range @xmath206 , where the functions @xmath89 and @xmath82 identifying the configurations with anti - stiff and stiff shells are given by ( [ eq : fminus ] ) and ( [ eq : fplus ] ) . in case of the schwarzschild exterior geometry the maximal surface compactness @xmath207 is achieved with a stiff shell and flat geometry in the gravastar interior ( compactness in the interior side of the shell @xmath208 ) , i.e. in the limit of vanishing ( dark ) energy density in the interior . introducing the ( dark ) energy density in the interior ( @xmath209 ) does not lead to higher surface compactness that can be reached .
this finding may be seen as counter - intuitive since one might have expected that the repulsive character of the de sitter geometry will help to support the shell against the collapse .
however , as our analysis has shown , this is not the case if the shell is to satisfy the dec . for completeness
, we have extended our analysis to the case of the schwarzschild de sitter exterior geometry which places the gravastar model in the context where the positive cosmological constant @xmath132 is present in the einstein equations .
relevant bounds on the surface compactness have been obtained for all values of the configuration variable @xmath210 , @xmath177 being the gravastar radius . for @xmath211
the highest surface compactness is bounded below unity and is achieved with the stiff shell . at @xmath137 there is a wide range of configurations where the surface compactness can approach unity arbitrarily close and it includes shells under surface pressure , dust shells and shells under surface tension . for @xmath164
the highest surface compactness is again bounded below unity and is achieved with anti - stiff shells .
we have also applied the procedure of ref .
@xcite to compute the quantity @xmath212 which in the context of perfect fluids corresponds to the speed of sound .
in order not to violate the principle of causality one normally requires that the sound speed does not exceed the speed of light .
this requirement has given still stronger upper bounds on the gravastar surface compactness as compared to the bounds derived from the energy conditions .
however , we must emphasize that the procedure is based on the assumption that the exotic matter comprising the gravastar shell obeys a barotropic equation of state of the form @xmath172 , on the assumption that the shell is stable against the radial perturbations , and most importantly on the assumption that the fluctuations in surface tension and surface energy density propagate along the shell at the speed given by the usual formula stated above .
since the full theory describing the gravastar shell is not available none of these assumptions can be taken for granted .
similar situation has been encountered in the context of the stability analysis of wormhole solutions in ref .
@xcite .
all the above results were derived using the @xmath0-shell gravastar model , but they may be relevant in the broader context of the dark energy stars , or gravastars with continuous pressure profiles , where one attempts to construct astrophysically plausible models of self - gravitating objects with dark - energy cores . as shown in ref .
@xcite , and applied in in ref .
@xcite , such objects require anisotropic principal pressures .
the analysis of the maximal surface compactness of anisotropic spheres is considerably more complicated than in the isotropic case ; probably the most general results obtained in this area is ref .
@xcite , although still not general enough to include the case of the dark energy interior , or ref .
@xcite where the influence of the cosmological constant on the compactness and other physical parameters of anisotropic configurations was considered .
however , thin shells ( soap bubbles ) appear to be the most efficient configurations in attempts to reach high surface compactness of anisotropic spherical bodies .
the maximal compactness of the ` shell around a black hole ' of ref .
@xcite , or that of our flat interior gravastar ( hollow shell ) , clearly saturates the general bound derived in ref .
@xcite , and similar suggestions have also been given in ref .
we acknowledge the support from the croatian ministry of science under the project 036 - 0982930 - 3144 .
si thanks for hospitality the university of vienna where part of this work was carried out .
10 url # 1#1urlprefix[2][]#2 mazur p o and mottola e 2001 ( _ preprint _ )
|
we consider the gravastar model where the vacuum phase transition between the de sitter interior and the schwarzschild or schwarzschild de sitter exterior geometries takes place at a single spherical @xmath0-shell .
we derive sharp analytic bounds on the surface compactness ( @xmath1 ) that follow from the requirement that the dominant energy condition ( dec ) holds at the shell . in the case of schwarzschild exterior ,
the highest surface compactness is achieved with the stiff shell in the limit of vanishing ( dark ) energy density in the interior . in the case of schwarzschild
de sitter exterior , in addition to the gravastar configurations with the shell under surface pressure , gravastar configurations with vanishing shell pressure ( dust shells ) , as well as configurations with the shell under surface tension , are allowed by the dec .
respective bounds on the surface compactness are derived for all cases .
we also consider the speed of sound on the shell as derived from the requirement that the shell is stable against the radial perturbations .
the causality requirement ( sound speed not exceeding that of light ) further restricts the space of allowed gravastar configurations .
|
the unusual and remarkable transport properties of graphene have attracted considerable attention , including : an unconventional quantum hall effect @xcite ; the possibility of testing the klein paradox @xcite ; the aharonov - bohm effect in graphene rings @xcite , as well as mesoscopic effects , such as weak localization @xcite , conductance fluctuations @xcite , quantum noise @xcite , coulomb blockade @xcite , and anderson localization @xcite ; specular andreev reflection and josephson effect @xcite ; formation of a wigner crystal @xcite ; voltage - driven quantum oscillations of the conductance @xcite ; intriguing electron lensing @xcite , and other fascinating phenomena ( see , e.g. , refs . and references therein ) .
studies of graphene are also inspired by its potential application in nanoelectronic devices , because the electron concentration can vary considerably due to applied electric fields , and graphene can have both electrons and holes as high - mobility charge carriers .
a main feature of the graphene electron structure , which is very different from conventional two - dimensional electron systems , is the existence of six dirac cones at the corners of an hexagon - shaped brillouin zone with a massless linear electron - hole dispersion .
this specific spectrum for the charge carriers leads to a number of interesting transport properties , or imparts new features to them .
the phenomena listed above are caused by the quantum peculiarities of graphene , and these manifest at the _ quantum _ level .
however , it is worthwhile to emphasize that the quantum features of the graphene conductivity can also play a very important role in _ classical _ macroscopic phenomena .
a nontrivial example of a classical phenomenon is the propagation of the electromagnetic waves localized near a graphene layer inserted into a dielectric @xcite .
it is known that surface electromagnetic waves can not propagate along the interface between two dielectrics .
surprisingly , the addition of only one monatomic graphene layer changes radically the _ macroscopic _ electrodynamic properties of the system .
the graphene layer can support a surface wave between two dielectrics and can play a very important role in other problems of plasmonics ( see , e.g. , refs . ) .
the dirac spectrum of electrons leads to new features of the electrodynamic response of the electron - hole plasma in graphene , as compared to conventional electron systems .
for instance , the existence of a localized transverse - electric ( te ) mode , a mode which can not exist in systems with a parabolic electron dispersion , was predicted in ref . for the graphene in
a _ symmetrical dielectric environment_. thus , both the transverse magnetic ( tm ) and transverse electric surface waves can propagate along the graphene layer . however , these tm and te modes exist in very different frequency ranges . indeed , as shown in ref .
, the frequency range for the te modes can vary from radio to infrared frequencies , depending on the carriers concentration , but the tm surface waves do not exist at the frequencies for the te modes . in this paper
, we predict the coexistence of localized both te- and tm - polarized electromagnetic waves that can propagate _ in the same frequency range _ along the graphene layer inserted into a photonic crystal ( pc ) .
it is important to emphasize that the te - polarized localized waves can propagate along the graphene layer placed between two semi - infinite dielectrics _ with very similar _ permittivities only .
here we consider the case when the graphene layer is in a _ non - symmetrical environment_. we show that , due to the periodic structure of the environment , localized te - polarized waves can exist even in such a non - symmetric case . moreover , this periodicity also allows the possibility for propagation of the te- and tm - polarized localized waves in the same frequency range , contrary to the case of propagation of the te- and tm - polarized waves along the graphene layers placed between two identical semi - infinite dielectrics .
we also consider the problem of the excitation of the te- and tm - polarized modes by the external wave that irradiates the pc - graphene - pc structure .
we show that the resonance peak of the transmissivity of the structure appears , when changing the frequency or the incident angle , due to the tm or te surface waves excitation , independently of the polarization of the exciting wave .
the predicted phenomenon can be observed in the terahertz frequency range , which is very important for various applications but not easily accessible with modern electronic and optical devices .
this technological perspective provides an additional motivation for study of these phenomena .
the analysis of the resonance peaks of the wave transmissivity can give important information on the graphene conductivity in the centimeter , millimeter , and submillimeter wavelength frequency ranges .
consider a graphene layer inserted into a photonic crystal ( pc ) .
the elementary cell consists of two nonmagnetic dielectrics with permittivities @xmath0 , @xmath1 and thicknesses @xmath2 , @xmath3 , respectively .
the period of the pc structure is @xmath4 .
the @xmath5-axis is perpendicular to the layers of the pc , and the graphene layer is arranged in the plane @xmath6 .
thus , the photonic crystal occupies the half - spaces @xmath7 and @xmath8 ( see fig . [ fig1 ] ) .
( color online ) geometry of the problem for waves localized on the graphene layer.,width=15,height=11 ] first we consider a tm - polarized surface wave with wave vector @xmath9 and with the following components of the electric @xmath10 and magnetic @xmath11 fields proportional to @xmath12 : @xmath13 and @xmath14 .
the fields in the photonic crystal satisfy the translation condition @xcite , @xmath15 and the bloch relation @xmath16 where @xmath17 is an integer number , @xmath18 is the propagation matrix for the tm wave with the elements , @xmath19 @xmath20 @xmath21 @xmath22 where @xmath23 ( @xmath24 ) , @xmath25 , @xmath26 , @xmath27 is the wave frequency , @xmath28 , @xmath29 is the complex bloch number , and @xmath30 @xcite . for the determinacy
, we assume that @xmath31 . in this case , the signs `` + '' and `` @xmath32 '' in the exponent in eq .
correspond to the electromagnetic fields in the regions @xmath8 and @xmath7 , respectively .
the boundary conditions on the graphene layer ( at @xmath6 ) consist of the continuity of the component @xmath33 of the electric field and the presence of the jump of the magnetic field component @xmath34 , caused by the current in graphene , @xmath35 here @xmath36 is the graphene conductivity , which is the sum of the intraband conductivity @xmath37 and interband conductivity @xmath38 @xcite . for a degenerate electron gas , when @xmath39 ( here @xmath40 is the boltzmann constant , @xmath41 is the graphene temperature , and @xmath42 is the chemical potential ) , @xmath37 and @xmath38 can be written as @xcite , @xmath43 @xmath44,\\ ] ] where @xmath45 is the intraband electron relaxation frequency and @xmath46 is the heaviside step function . using eqs .
( [ eq1][eq9b ] ) , we obtain the following dispersion relation for the tm - polarized surface wave : @xmath47 similarly , we can derive the dispersion relation for the te - polarized surface wave with the following components of the electromagnetic field and the wave vector : @xmath48 , @xmath49 , and @xmath9 .
the fields in the photonic crystal satisfy the conditions @xcite , @xmath50 @xmath51 where @xmath52 is the propagation matrix for the te wave with components : @xmath53 @xmath54 @xmath55 @xmath56 afterwards , we obtain the dispersion relation for the te surface wave in the form , @xmath57 where @xmath58 @xcite .
for the numerical calculations of the dispersion curves , we assume that @xmath59 k and the graphene electron concentration @xmath60 m@xmath61 .
the corresponding chemical potential @xmath42 is about 45 mev ( or 522 k ) .
these parameters are in agreement with experimental results @xcite .
we will neglect the dissipative losses and consider the graphene conductivities @xmath37 and @xmath38 as purely imaginary values .
curves 13 in fig .
[ fig2 ] show the numerically - calculated dispersion curves for the tm - polarized surface waves .
the wave frequency is normalized to @xmath62 sec@xmath63 ( terahertz range ) .
hereafter , we assume the following parameters for the photonic crystal : @xmath64 , @xmath65 ( this value corresponds to silica glass ) , and @xmath66 ( teflon ) ( see refs .
curves 4 and 5 are the light lines for silica glass and teflon , respectively .
( color online ) dispersion curves ( 13 ) for the tm - polarized surface waves in the pc - graphene - pc structure in the thz frequency range .
curves 4 and 5 are the light lines for silica glass and teflon , respectively . here
the forbidden zones are marked in yellow .
the square on curve 2 indicates the values of @xmath27 and @xmath68 which correspond to the peak of the transmissivity arising due to the excitation of the tm - polarized mode ( see the next section).,width=15,height=11 ] all the dispersion curves run within the forbidden zones ( regions marked in yellow ) of the infinite photonic crystal .
note that the edges of these forbidden zones , expressed in dimensionless coordinates @xmath69 and @xmath70 , are the same for different frequency ranges .
however , the positions of the dispersion curves depend substantially on the choice of @xmath71 .
one can see that each of the forbidden zones of the photonic crystal contains one branch of the spectrum for the tm - polarized surface wave .
note that the dispersion curve , which corresponds to the dispersion relation for the waves localized near the graphene layer placed between the semi - infinite teflon and semi - infinite silica glass , practically coincides with curve 1 in fig .
this coincidence means that replacing the semi - infinite teflon and semi - infinite silica glass by the photonic crystals does not change significantly the dispersion curve for the tm - polarized surface wave in the thz frequency range .
however , this remark only concerns the first forbidden zone .
this fact has a simple physical interpretation .
namely , the plasmons ( tm modes ) near the graphene layer are extremely localized and , at large frequencies , the evanescent tail can be smaller than the thickness of the dielectric layers .
in other words , graphene can truly `` see '' only the nearest uniform dielectrics . on the other hand , in the low - frequency limit
( when the evanescent tail becomes larger ) , the graphene layer will be able to `` see '' many layers of the photonic crystal , changing the plasmon dispersion relation . to demonstrate this , we present some numerical estimates for the dimensionless localization depth @xmath72 of the electromagnetic fields of the tm - polarized waves . recall that @xmath68 is the wave vector of the localized mode and @xmath73 is the bloch decay constant .
physically , the value of @xmath74 is the ratio of the distance from the graphene layer , where the field amplitudes are reduced by a factor of e @xmath75 , to the wavelength . for the mode in the first forbidden zone
, we have : @xmath76 in the thz frequency range and @xmath77 in the centimeter - wavelength frequency range ( for @xmath78 sec@xmath63 ) .
note that curves 2 and 3 on fig .
[ fig2 ] seem to be lying precisely at the boundaries of the forbidden regions ( marked in yellow ) .
this means that these modes are weakly localized near the graphene layer
. indeed , we have @xmath79 , and @xmath80 for the modes in the second and third forbidden zones , respectively .
the main new remarkable feature of the spectrum for the surface waves in the pc - graphene - pc system , in comparison with the teflon - graphene - silica glass structure , consists in the appearance of additional branches of the spectrum in the second , third , and so on , forbidden zones .
curves @xmath81 , @xmath82 in fig .
[ fig3 ] show the dispersion curves for the te - polarized surface waves in the thz frequency range , for the same parameters of the photonic crystal as in fig .
curves 3 and 4 in fig .
[ fig3 ] are the light lines for silica glass and teflon , respectively .
the principal difference of the spectrum for the te - polarized waves , in comparison with tm waves , is the absence of dispersion curves in the first forbidden zone for the terahertz range .
indeed , the dispersion curve appears in the first forbidden zone only when @xmath83 .
in contrast to the usual two - dimensional electron gas with drude conductivity , this inequality can be satisfied in monolayer ( or bilayer @xcite ) graphene .
for the chosen values of @xmath84 and @xmath41 , this condition is only satisfied starting with the infrared ( ir ) frequencies . as seen from fig .
[ fig3 ] , the dispersion curves for the te modes , as well as for the tm modes ( see the dispersion curves @xmath82 and @xmath85 in fig . [ fig2 ] ) , are very close to the boundaries of the forbidden zones .
this means that these modes are weakly localized near the graphene layer .
the numerical estimates of the dimensionless localization depth @xmath74 for the te waves give similar values as for the corresponding tm - polarized modes .
( color online ) dispersion curves for the te - polarized surface waves in the structure pc - graphene - pc in the thz frequency range ( curves 1 and 2 ) .
curves 3 and 4 are the light lines for silica glass and teflon , respectively . here
the forbidden zones are marked in yellow .
the square on curve 1 indicates the values of @xmath27 and @xmath68 which correspond to the peak of the transmissivity arising due to the excitation of the te - polarized mode ( see the next section).,width=15,height=11 ] now we consider the dispersion curves of the te - polarized modes in the ir frequency range , where unusual transport properties of the graphene layer manifest themselves .
these curves ( 1 and 2 ) are shown in fig .
[ fig4 ] for the same parameters of the photonic crystal as in fig .
[ fig3 ] ( however , with the dielectric permittivities @xmath86 and @xmath87 for silica glass and teflon in the ir frequency range , see refs .
the wave frequency is normalized to @xmath88 sec@xmath63 .
( color online ) dispersion curves for the te - polarized surface waves in the structure pc - graphene - pc in the ir frequency range ( curves @xmath81 and @xmath82 ) .
curves 3 and 4 are the light lines for silica glass and teflon , respectively . here
the forbidden zones are marked in yellow .
the open circle indicates the starting point of dispersion curve 1 .
the part of curve @xmath81 between the circle and triangle corresponds to the highest degree of localization of the mode s field.,width=15,height=11 ] as seen from fig .
[ fig4 ] , dispersion curve 2 for the te - polarized mode in the second forbidden zone is very close to the boundary of the forbidden zone , similarly to the case of the thz frequency range .
this means that this mode is also weakly localized near the graphene layer .
the dispersion curve for the te mode in the first forbidden zone exists only in a finite - frequency interval .
the edges of the corresponding part of the dispersion curve are depicted by the open circle and open triangle on curve 1 in fig .
[ fig4 ] . at higher frequencies , the bloch phase @xmath90 in the dispersion relation given by eq .
is very small , @xmath91 .
the localization depth is much larger than the wavelength of the te mode in this case . to clarify this , it is worthwhile to analyze the frequency - dependence of the conductivity , @xmath92 , for the given values of @xmath42 , and @xmath41 , at @xmath93
. this dependence is shown in fig .
[ fig5 ] for the ir frequency range . the open circle and open triangle in fig .
[ fig5 ] are in agreement with the same symbols shown on curve @xmath81 in fig .
[ fig4 ] .
( color online ) frequency dependence of the dimensionless conductivities @xmath94 ( curve 1 ) , @xmath95 ( curve 2 ) , and @xmath96 ( curve 3 ) calculated by means of eqs .
and for the given values of @xmath97 k , @xmath59 k , and @xmath88 sec@xmath63 , at @xmath93.,width=15,height=11 ] the maximum absolute value of @xmath98 corresponds to the maximum of the bloch phase @xmath99 and to the strongest localization of the electromagnetic field of the te wave near the graphene layer .
the conductivity @xmath98 tends to zero when increasing the frequency . as a result
, the localization depth becomes 100 times larger for the point marked by the open triangle on curve @xmath81 in fig .
[ fig4 ] .
note that there are three necessary conditions for the existence of te - polarized surface waves propagating along a conducting layer placed between _ semi - infinite homogeneous _ dielectrics .
first , according to ref . , the linear electron - hole dispersion law , specific for the graphene layer , should be valid .
second , the condition @xmath83 should be satisfied . finally , the permittivities of the two dielectrics surrounding the graphene layer should be _ very similar_. here we now consider the system pc - graphene - pc where the graphene layer is placed between two _ very different _ adjacent dielectrics .
so , at first glance , the localized te modes can not exist in such a system because the third condition mentioned above is not satisfied .
however , the dimensionless localization depth @xmath74 for all the te surface modes considered are extremely large , and the graphene layer can actually `` see '' a symmetric dielectric environment which enables the existence of the te modes . here
we present results of our numerical analysis of the dispersion curve for the te localized mode in the lowest forbidden zone of the photonic crystal versus the thicknesses @xmath2 and @xmath3 of the dielectric layers . in the case
@xmath100 , i.e. , when the thicknesses of the silica glass layers are greater than the thicknesses of the teflon layers , the boundary of the first forbidden zone of the photonic crystal in fig .
[ fig4 ] , as well as the dispersion curve for the te - polarized mode ( curve @xmath81 in [ fig4 ] ) shift towards the light line in silica glass ( curve 3 in fig .
[ fig4 ] ) .
when the inverse inequality is realized , i.e. , when @xmath101 , the boundary of the first forbidden zone and the dispersion curve for the te - polarized mode shift towards the light line in teflon ( curve 4 in fig .
[ fig4 ] ) .
in addition , the dispersion curve for the te - polarized mode shifts towards the light line in silica glass when increasing the thicknesses of the pc layers ( for @xmath102 ) and tends to this light line in the limit @xmath103 .
this means that , in the limit @xmath103 , the electromagnetic field of the localized te wave becomes delocalized in silica glass , and the localized wave itself does not exist . in this section
, we used the floquet theorem @xcite to derive the dispersion relations for the tm- and te - polarized eigenmodes localized near the graphene layer inserted into a photonic crystal with an infinite number of layers .
as known , this theorem is valid for the non - dissipative case only .
however , the behavior of the dispersion curves does not change significantly for systems with small enough dissipative parameters .
we demonstrate this in the next section where we study the problem of the excitation of localized waves in the pc - graphene - pc structure with a finite number of layers .
we will not use the floquet theorem , but instead apply the propagation - matrix method , which allows us to take into account realistic dissipation parameters .
in this section , we study the transmittance of the pc - graphene - pc structure for te - polarized waves .
the calculation of the transmission coefficient for the tm - polarized waves is performed in the same way .
we now consider the excitation of the te waves propagating along the graphene layer placed in the middle of the photonic crystal containing a _ finite number _
@xmath104 of elementary cells .
this structure is irradiated , by an electromagnetic wave , from its right side .
the external wave is incident from a dielectric with permittivity @xmath105 on the pc - graphene - pc structure under the angle @xmath106 , passes through the pc - graphene - pc structure , and , finally , transmits to another external dielectric with the same permittivity @xmath105 .
evidently , the resonance excitation of the surface wave will occur if its polarization , frequency , and the component @xmath68 of the wave vector coincide with the parameters of the incident wave .
we define the electric and magnetic fields in the dielectric on the right - hand side of the structure ( at @xmath107 ) as @xmath108 @xmath109,\\end{aligned}\ ] ] where @xmath110 and @xmath111 are the amplitudes of the incident and reflected waves , respectively .
we define the electromagnetic field of the wave on the left - hand side of the structure ( at @xmath7 ) as @xmath112 where @xmath113 is the amplitude of the transmitted wave . to find the transmission coefficient @xmath114 , we use the condition of the magnetic field discontinuity on a graphene layer ( similar to eq . written for the tm wave ) and eq .
following the procedure described in ref .
@xcite , we derive , @xmath115 with @xmath116 . substituting eqs .
( 18)([eqs3 ] ) into eq . , we derive the following relation for the wave amplitudes : @xmath117 solving eq .
, we have the following expression for the fresnel transmissivity coefficient : @xmath118 with @xmath119 .
the corresponding expression for the tm - polarized wave is @xmath120 here @xmath121 and @xmath122 are the elements of the matrices @xmath123 and @xmath124 .
the matrices @xmath125 and @xmath126 are , @xmath127 we now study the frequency dependence @xmath128 of the transmissivity of the pc - graphene - pc structure , which should have a sharp maximum when the resonance excitation of the surface wave takes place .
figure [ fig6 ] shows the frequency dependence @xmath129 for the tm- and te - polarized waves at @xmath62 sec@xmath63 , @xmath130 , @xmath131 , @xmath132 ( vacuum ) , and @xmath133 sec@xmath63 .
the chosen value of the relaxation frequency is in agreement with the theoretical and experimental results @xcite .
( color online ) frequency dependence of the transmissivity @xmath129 for the pc - graphene - pc structure for the te ( curve a ) and tm ( curve b ) waves .
the resonance frequencies and the values of other parameters correspond to the squares on the dispersion curves in figs . 2 and 3 .
note that we have @xmath134 and @xmath135 for the maxima on curves a and b , respectively.,width=15,height=11 ] the resonance peaks of the transmissivity are observed within the forbidden zone at frequencies that correspond to the open squares on the dispersion curves in figs .
[ fig2 ] and [ fig3 ] .
note that these peaks are clearly seen in spite of the weak localization of the excited modes .
the positions of the peaks , that can be measured for any wave polarization , and the quite - simple eqs . and allow one to calculate the graphene conductivity in the centimeter , millimeter , and submillimeter wavelength frequency ranges . note also that , in the absence of the graphene layer in pc , the peaks of the transmissivity are not observed .
the presence of a graphene layer within a photonic crystal results in a concentration of the electromagnetic field near it . in this section ,
we discuss the spatial distribution of the squared amplitudes @xmath136 and @xmath137 of the electric and magnetic fields in the problem of the excitation of the localized wave by the incident electromagnetic wave ( see the previous section ) .
the distributions of @xmath138 and @xmath139 for the tm - polarized wave are qualitatively the same as the distributions @xmath136 and @xmath137 for the te wave . for numerical calculations , we chose the frequency which corresponds to the maximum of the peak on curve a in fig .
the results of the calculations are shown in figs .
[ fig7 ] and [ fig8 ] .
the values @xmath136 and @xmath137 are normalized to the corresponding values of the incident wave .
the dashed vertical lines in the figures show the position of the graphene layer .
we assume that the light is incident from the right - hand side of the structure , i.e. , the wave propagates from the unit cell with number 30 to the unit cell with number 1 .
( color online ) the distribution of the squared amplitude @xmath136 of the dimensionless electric field for the te - polarized wave @xmath140 in the pc - graphene - pc structure.,width=15,height=11 ] ( color online ) the distribution of the squared amplitude @xmath137 of the dimensionless magnetic field for the te - polarized wave @xmath141 in the pc - graphene - pc structure.,width=15,height=11 ] figures [ fig7 ] and [ fig8 ] show that the maxima in the distributions of the electric and magnetic fields are near the graphene plane .
in addition , the value of @xmath137 suffers a discontinuity on the graphene plane ( see the inset in fig .
[ fig8 ] ) which is caused by the ac current excited in the graphene layer by the incident wave .
to conclude , we predict the coexistence of localized te- and tm - polarized surface electromagnetic waves in a pc - graphene - pc structure .
these waves can propagate _ in the same frequency range _ due to the zone structure of the photonic crystal , in contrast to the waves localized near the graphene layer placed between two identical dielectrics . in the latter case , the tm and te surface waves can propagate in different frequency ranges .
we also consider the excitation of localized tm and te modes by the external wave that irradiates the pc - graphene - pc structure .
we show that , independently of the polarization of the exciting wave , the resonance peak of the transmissivity of the structure appears when changing the frequency or the incident angle .
the analysis of the resonance peaks of the wave transmissivity can provide important information on the graphene conductivity in the centimeter , millimeter , and submillimeter wavelength frequency ranges .
we acknowledge partial support from the riken ithes project , jsps - rfbr contract no . 12 - 02 - 92100 , grant - in - aid for scientific research ( s ) , and the program fpnnn of the nas of ukraine ( project no .
0110u005642 ) .
novoselov , a.k .
geim , s.v .
morozov , d. jiang , m.i .
katsnelson , i.v .
grigorieva , s.v .
dubonos , and a.a .
firsov , nature * 438 * , 197 ( 2005 ) ; y. zhang , y .- w .
tan , h.l .
stormer , and p. kim , nature * 438 * , 201 ( 2005 ) .
|
we predict the coexistence of both te- and tm - polarized localized electromagnetic waves that can propagate _ in the same frequency range _ along a graphene layer inserted in a photonic crystal .
in addition , we studied the excitation of these modes by an external wave and have shown that the resonance peaks of the sample transmissivity should be observed due to the excitation of the localized waves , independently of the polarization of the exciting wave .
the simplicity of the derived dispersion relations for the localized modes and the possibility to excite waves of both polarizations provide a method for measuring graphene conductivity .
|
incoming solar radiation at the top of the atmosphere is an important quantity in many areas of earths systems modeling .
this physical quantity is needed in areas ranging from low dimensional energy balance models ( e.g. the budyko energy balance model @xcite ) to large global circulation models , gcms , ( e.g. nasa s modele ar5 @xcite ) .
it is common practice to compute insolation by latitude using computer algorithms .
for example , nasa s latitudinal insolation calculations for modele ar5 rely on three fortran subroutines that 1 ) calculate earth s orbital parameters ( eccentricity , obliquity , and longitude of perihelion ) as a function of year , 2 ) calculate distance to the sun and declination angle as functions of time of year and orbital parameters , and 3 ) calculate the time integrated zentih angle as a function of the declination angle and the time interval of the day @xcite .
these computer calculations are useful if you are working with a grid version of a planet ( as is typical in gcm s ) , however , to convert this information to useable data for other modeling scenarios is not always straightforward .
for example , in the budyko - widiasih energy balance model , one must know the insolation as a function of latitude in order to make use of the model @xcite .
fitting a polynomial , trigonometric function , or spline to data points given by a computer program in order to obtain such a function is unideal because it obscures the true relationship between insolation and latitude . in the following section we highlight a method developed by mcgehee and lehman in @xcite which gives the mean annual insolation by latitude for any planet as a function of obliquity and eccentricity . in the last section
we discuss how the results of this method compare to results obtained from two computer simulations , one for earth @xcite and the other for pluto @xcite .
in `` a paleoclimate model of ice albedo feedback forced by variations in earth s orbit '' mcgehee and lehman develop a method to compute the mean annual insolation by latitude using only mathematical principles ( section 5 in @xcite ) .
they found that one can express mean annual insolation , @xmath1 , as a function of eccentricity @xmath2 , obliquity @xmath0 , and sine of latitude @xmath3 by finding the insolation at any point on the earth s surface , integrating over the course of one orbital period , then integrating over all longitudes @xcite .
their results are @xmath4 where the distribution of insolation across the sine of the latitude is given by @xmath5 ( where @xmath6 is longitude ) and the magnitude of insolation is given by @xmath7\frac{1}{\sqrt{1-e^2}}. \label{eq : global}\ ] ] where @xmath8 is the semi - major axis , @xmath9 is the solar output in watts , @xmath10 is the mass of the sun , and @xmath11 is the mass of earth .
we see that their analysis is general enough to apply to any planet orbiting a star as long as the appropriate physical parameters are known .
it is important to note that mcgehee and lehmen account for axial precession in their initial analysis .
the precession dependence for annual insolation averages out over the course of a year because of their assumption that precession is constant over this time period . earth s precession period is around 26,000 years which means that the precession angle changes by about @xmath12 radians each year .
this small change in precession angle is negligible for the earth .
it is important to note that other planets axial precession may exhibit resonances with the planet s orbital period , resulting in non - negligible precession .
also , note that the distribution function ( equation [ eq : distribution ] ) is symmetric in @xmath3 .
we have @xmath13 with the change of variable @xmath14 we get : @xmath15\right)^2}\ , d\tilde\gamma\end{aligned}\ ] ] and with another change of variable @xmath16 and using the periodicity of sine we get the desired result : @xmath17\right)^2}\ , d\alpha\\ & = \frac{2}{\pi^2}\int_{0}^{2\pi}\sqrt{1-\left(\sqrt{1-y^2}\sin\beta\sin(\alpha-2\pi)-y\cos\beta\right)^2}\ , d\alpha\\ & = s(y,\beta)\end{aligned}\ ] ] this means that as long as the obliquity , precession , or eccentricity values change very little over the course of a year , the mean annual insolation will be symmetric across the @xmath3-axis ( the planet s equator ) no mater what the values happen to be ..
in the following section we show how the method for section 2 compares with the results from two different computer programs designed to calculate insolation by latitude .
we first compare it with nasa s modele ar5 @xcite which has a program to compute insolation for the earth and then we compare our method with a program that computes the insolation for pluto @xcite . from equations [ eq : global ] and [ eq : distribution ] above and the values of eccentricity and obliquity from the nasa fact sheet for earth @xcite , we can get the annual mean insolation by latitude for 2013 ce .
we plot these results in figure [ fig : earth ] ( a ) along with data from the insolation calculations of modele ar5 . from the figure we can see that our results match nasa s results for the year 2013 almost identically .
we also looked at nasa s results for the year 4500 bce .
we chose this year because earth s period of precession is about 26,000 years and 4500 bce is a quarter of this period .
we used laskar s calculations for obliquity and eccentricity , as these parameters change slowly over time and the values for 6,000 years ago are different from the current values .
we take eccentricity to be @xmath18 and obliquity to be @xmath19 @xcite .
note that @xmath0 is given in radians . from figure
[ fig : earth ] ( b ) we see that our results coincide with nasa s modele results again . from equations [ eq : global ] and [ eq : distribution ] above and the values of eccentricity and obliquity from the nasa fact sheet for pluto @xcite , we can get the annual mean insolation by latitude .
we plot these results in figure [ fig : pluto ] ( a ) along with the results of @xcite ( figure [ fig : pluto ] ( b ) ) . from the figure we can see that our results differ both quantitatively and qualitatively and .
note that we are referencing only the part of earle and bindles data that corresponds to the annual average by latitude , which is the green circle plot in figure [ fig : pluto ] ( b ) .
quantitatively , the two plots differ by a constant of @xmath20 , but are otherwise share the same order of magnitude . on the qualitative side , we see that earle and binzel s results are slightly asymmetric , indicating that in pluto s most recent year ( since 1767 ) its north pole has received more insolation than its south pole .
they claim that this slight asymmetry is due to the fact pluto s longitude of perihelion is small ( about @xmath21 ) , resulting in pluto s line of equinox s almost aligning with its perihelion .
recall that our formula for insolation does nt depend on the precession angle .
note that the longitude of perihelion is determined by the precession angle ( as the planet precesses , its line of equinoxes changes ) , so our results do nt depend on the longitude of perihelion .
one area of concern could be that pluto s period of precession is in resonance with its orbital period .
dobrovolskis et al .
show that the angle between pluto s perihelion and its vernal equinox have a period of about three million earth years , or about 12,000 pluto years @xcite .
although this period is slightly faster than earth s precession , it is large enough so that pluto s precession is negligible in a pluto year . thus , we should have no influence from the precession angle ( or longitude of precession ) in the calculations of pluto s insolation .
|
common methods for calculating a planet s annual insolation by latitude have relied on computationally heavy or complex computer algorithms . in this paper , we show that mean annual insolation by latitude of a planet with obliquity angle @xmath0 can be found by taking the definite integral of a function of longitude . this leads to faster computations and more accurate results .
we discuss differences between our method and selected computational results for insolation found in the literature .
|
closely separated qso pairs and gravitationally lensed qsos provide two or more adjacent lines of sight ( los ) , which allow to sample the size and clustering of the absorbers .
the ray path separation of gravitationally lensed qsos usually spans subgalactic scales ; studies of these spectra have shown that the ly@xmath0 clouds are much larger than a few kiloparsecs ( weymann & foltz 1983 ; foltz et al .
1984 ; smette et al .
1992 , 1995 ) .
further results again pointing at very large sizes have been obtained for los separations of a few arcminutes or less ( shaver & robertson 1983 , sr83 ; crotts 1989 ; dinshaw et al .
1995 ; dinshaw et al . 1997 , d97 ; fang et al . 1996 , fdcb ; crotts & fang 1997 , cf97 ) .
the first indication in this sense has been obtained observing the qso pair q1343 + 2640a , b ( with a proper sightline separation of @xmath11 kpc at @xmath12 ) , indicating a cloud radius of @xmath13 kpc ( bechtold et al .
1994 ; dinshaw et al .
qso pairs with separations larger than a few arcminutes , corresponding to proper separations up to megaparsecs , have the potential to study the large - scale spatial distribution of ly@xmath0 or metal absorbers ( crotts 1985 ; jakobsen et al .
1986 ; tytler , sandoval & fan 1993 ; elowitz , green & impey 1995 ; dinshaw & impey 1996 ; williger et al .
1996 ) .
multiple sightlines are also highly valuable in probing the large scale structure .
in particular , multiple , well - sampled sightlines can provide a stronger test for voids by placing more absorbers in a void - sized volume than could possibly be obtained along a single sightline .
furthermore , the effect of the uv radiation field of a foreground qso on the absorption spectrum of a background qso provides an important way of testing the `` proximity effect '' interpretation ( bajtlik , duncan & ostriker 1988 ) of the `` inverse effect '' observed in single qso sightlines .
we present new observations of the qso pair q0307 - 195a , b discovered in an objective prism survey by macalpine and feldman ( @xcite ) ( a = and b = ) .
later on , the pair was observed by shaver and robertson ( @xcite ) who analyzed the absorption spectra of the two objects .
both qsos are about of 19th magnitude , their redshifts are @xmath14 and @xmath15 ( sr83 ) , and they are separated by 56 arcseconds on the plane of the sky . the redshift difference is small enough ( @xmath16 km s@xmath2 ) that they could be members of the same cluster . the rest of the paper is organized as follows : in 2 , we present spectroscopic observations of q0307 - 195a , b and describe the procedure of reduction and calibration of the spectra . in 3 the metallic systems in the two qso spectra are listed and briefly discussed . section 4 describes the procedure to obtain an estimate for the radius of the ly@xmath0 absorbers in the hypothesis of simple geometry . in subsection 4.1
we tackle the same issue on the basis of more realistic models for the absorbers . finally , in 5 we summarize our results .
-6 cm -6 cm we obtained spectra of the qso pair q0307 - 195a , b with the blue arm of the emmi spectrograph at the 3.5 m new technology telescope of the european southern observatory in two runs in november 1994 and november 1995 .
the spectrograph was used with the holographic 3000 lines mm@xmath2 grating ( eso # 11 ) and a @xmath17 , 24 @xmath18 m pixel , tektronix ccd ( eso # 31 ) .
two or three spectra were obtained at each of three different central wavelengths : 3690,3750 and 3850 ( ly@xmath0 forest region ) and one at 4780 ( region ) .
typical exposure times were 6000 s / spectrum for a total exposure time of 56900 s. in this instrument configuration one ccd pixel corresponds on average to 0.15 and 0.13 in the uv and blue regions respectively .
the typical slit width was 1.2 arcsec in width and 180 arcsec in length and it was aligned at p.a .
@xmath19 to capture both qsos .
the spectra were extracted , wavelength calibrated , normalized and merged using standard procedures of the eso midas _ long slit _
reduction package .
the final smoothed and merged spectra cover the ranges @xmath20 and @xmath21 .
the resolution , as measured from the thorium lines of the calibration spectra , extracted and treated in the same way as the qsos spectra , is 0.7 and 0.5 in the uv and blue range respectively .
the @xmath22 ratio of the final spectra varies from 25 at the peak of the ly@xmath0 emissions of the two qsos to 5 at the the shortest wavelengths .
the relative accuracy of the wavelength scales in the spectra of the two qsos is better than 20 km s@xmath2 .
the normalized spectra are shown in figs .
[ spec1 ] , [ spec2 ] .
[ t1 ] [ t2 ]
the determination of the continuum in the qso spectrum is a critical step because it affects the measurement of the absorption line parameters . in the present work , we selected the portions of the spectrum free of strong absorption lines or artificial peaks ( e.g. due to cosmic rays ) , i.e. where the rms fluctuation about the mean becomes consistent with noise statistics , and then , we estimated the continuum level by spline - fitting these regions with quadratic polynomials . in order to detect the absorption lines we have searched for all the features deviating from the continuum more than 3.5 times the rms noise .
the lines have been fitted with voigt profiles convolved with the instrumental spread function , making use of a minimization method of @xmath23 .
this step has been performed within the eso midas _
lyman _ package ( fontana & ballester 1995 ) .
the values of the wavelength , redshift @xmath24 , and rest equivalent width @xmath25 have been determined for isolated lines and individual components of blends .
the number of components of each absorption feature is assumed to be the minimum required to give a reduced @xmath26 ( corresponding to a confidence level @xmath27 ) .
the error on the observed equivalent widths was computed in such a way that it includes also the uncertainty in the determination of the level of the continuum and in the choice of the fitting interval for the lines . due to the improved signal to noise ratio ( @xmath22 ) and resolution
, we observe more than twice the number of lines found by sr83 in the common wavelength interval .
the lists of lines are presented in the tables [ t1 ] , [ t2 ] .
we based the identification of metal lines on sr83 .
all the systems reported by them in the common wavelength interval are confirmed , in addition we found one new system in q0307 - 195a and one possible new system in q0307 - 195b ( see 3.2 and 3.3 ) .
an identification programme , based on the method of young et al .
( 1979 ) , has been applied to the newly observed lines that could not be attributed to systems already identified by sr83 .
because of the limited wavelength coverage in the red region of the spectra , we were able to identify only one new doublet .
four and five lines redward of the ly@xmath0 remained unidentified in q0307 - 195a and b respectively .
the metal systems of each object are described in detail in the following section .
: : _ the metal system at @xmath28 - a _ + already observed by sr83 .
this absorption system presents a redshift difference of only 115 km s@xmath2 with object b and of 77 km s@xmath2 with system _ a _ of object b. sr83 argued that this common absorption is likely due to a very extended gaseous halo or disc , physically associated with q0307 - 195b itself .
: : _ the metal system at @xmath29 - b _ + this is a new system .
we identified the and doublets in the red portion of the spectrum , and the ly@xmath0 line blended with the 1207 of the previous system .
: : _ the metal systems at @xmath30 - c and @xmath31 - d
_ + already observed by sr83 . no new line has been added .
: : _ the metal systems at @xmath32 - a and @xmath33 - b
_ + already observed by sr83 .
no new line has been added .
: : _ the metal system at @xmath34 - c
_ + already observed by sr83 .
the identification is based only on the doublet , in the ly@xmath0 forest a possible can be identified but it is blended with a strong ly@xmath0 line .
the doublet shows velocity structure and it was fitted with two components .
: : _ the metal system at @xmath35 - d _
+ the identification of this possible new system is based only on the doublet found redward of the ly@xmath0 emission line .
this system differs in redshift by only 166 km s@xmath2 from system _ d _ in object a. all the lines in the ly@xmath0 forest not identified as metals are assumed to be ly@xmath0 .
[ thit ] -9 cm -11.5 cm [ t3 ] -1 cm = 15.2 cm -12.5 cm the main aim of this work is to determine the transverse size of the ly@xmath0 absorbers using the ly@xmath0 forests of the two adjacent loss of the pair . in order to avoid spurious effects due to the `` proximity effect '' ,
only the ly@xmath0 lines with velocity separation greater than 5000 km s@xmath2 from the lower qso redshift in the pair were considered in the sample .
the spectra of the observed pairs show a number of _ hits _ ,
@xmath36 , and of _ misses _ , @xmath37 .
a _ hit _ occurs when an absorption line above a given threshold in rest equivalent width , @xmath25 , appears in both qso spectra , with a velocity difference less than a limit , @xmath38 .
a _ miss _ occurs when a line is seen in any of the qso spectra , but no line above the @xmath25 threshold is seen in the other .
a value @xmath39 km s@xmath2 has been adopted as the most suitable velocity window , being significantly larger than the resolution ( @xmath40 km s@xmath2 ) and close to the clustering scale observed for ly@xmath0 absorbers ( cristiani et al .
1997 ) . as a consequence ,
all the ly@xmath0 lines in the sample whose separation was less than @xmath41 km s@xmath2 have been merged into a single line with wavelength given by the average of the wavelengths and equivalent width equal to the sum of the equivalent widths of the `` parent '' lines . in table
[ thit ] , the hits observed in the spectra of q0307 - 195a , b are reported together with the velocity differences between the lines . to constrain cloud sizes ,
it can be assumed that a binomial random process produces the observed number of hits and misses ; if the probability for a hit is given by @xmath42 , the likelihood function of this binomial process is @xmath43 ( fdcb ) .
@xmath42 is the probability that both ray paths intersect the cloud , given that at least one ray path does .
@xmath42 is related to the probability @xmath44 , that one given los intersects the cloud given that the other adjacent ray path already does , by the formula : @xmath45 the functional form of the probability @xmath44 depends on the assumptions for the geometrical shape of the absorbers . in the hypothesis of single radius ,
spherical clouds , with the definition @xmath46 , where @xmath47 is the proper separation of ray paths and @xmath48 is the cloud radius , it is found ( mcgill 1990 ) : @xmath49\ \ \ \ \ \mathrm { for}\ x<1,\ ] ] and @xmath50 otherwise . as a second model , clouds can be idealized as circular discs , with a given radius @xmath48 , much larger then their thickness , and an observed inclination angle @xmath51 .
the probability that one ray path intersects the disc within the angle @xmath52 , given that the other ray path already does , is ( mcgill 1990 ) : @xmath53^{1/2}\bigg\ } \ \ \ \ \mathrm { for}\ x<\cos \theta,\end{aligned}\ ] ] and @xmath50 otherwise . by integrating over @xmath51 for randomly oriented discs , the probability is then given by : @xmath54 in practice , from the observations we determine the numbers of hits and misses , @xmath36 and @xmath37 , and then we correct them for the accidental hits due to random velocity matches .
two different procedures have been devised to compute the value of @xmath36 : in the first , the velocity separation between all the line pairs is computed and , when a given line combines to give more than one hit , only the one with the smallest velocity separation is taken into account , and the others are neglected . if , for example , lines 1 and 2 of spectrum a combine with lines 1 and 2 of spectrum b forming the couples 1 - 1 , 1 - 2 and 2 - 2 , and the couple 1 - 2 has the smallest velocity separation , we count just one hit .
the second method takes the minimum value between the `` a vs. b '' hits and the `` b vs. a '' hits .
the two methods turned out to yield the same result in all the investigated cases . the number of observed misses is given by the sum of the two @xmath37 s , `` a vs. b '' and `` b vs. a '' . to take into account the possibility of _ accidental hits _ due to random velocity matches , we have used simulations of poissonian distributions of absorption lines in the two qso loss , with a number of lines in each spectrum equal to the observed one . for each simulation
we have computed : - : : the number of accidental hits , @xmath55 , applying the same procedures used for the real spectra ; - : : the corrected number of hits , @xmath56 , ( misses , @xmath57 ) by subtracting @xmath55 from ( adding @xmath58 to ) the observed value . eventually , the probability distribution of the number of corrected hits @xmath59 was evaluated .
the probability @xmath60 takes into account all the occurrences with @xmath61 .
= 90truemm for each pair of values of corrected hits and misses , given the expressions for the probability distributions @xmath44 and @xmath42 , it is possible to apply bayes s theorem ( press 1989 , fdcb ) to yield the a posteriori probability density for the radius @xmath48 , @xmath62 at variance with fdcb , we have chosen as a prior distribution , @xmath63 , a uniform distribution for @xmath48 between 0 and @xmath64 , and not a uniform distribution of @xmath65 .
it is worth stressing that a uniform prior distribution for the probability @xmath42 introduces a cutoff at high and low @xmath48 that forces the 95 % confidence intervals to be narrow also in situations with a very low number of hits or misses . in particular , a spurious dependence of @xmath48 on the proper separation @xmath47 is induced in the case of very few hits , which is the common situation at large separations , because the probability @xmath65 actually depends on @xmath47 . since there is no particular justification to assume a uniform @xmath65 , we have chosen a uniform distribution for @xmath48 which does not introduce biases in the results .
= 90truemm = 90truemm equation ( 5 ) then becomes @xmath66 we have adopted as @xmath67 a value of @xmath68 mpc which represent the _ mean free path _ between two absorbers along the line of sight , as obtained from the line distribution @xmath69 ( giallongo et al .
1996 ) at a redshift @xmath12 , in the column density range @xmath70 . in the case of discs ,
the maximum value is increased by a factor @xmath71 , yielding a value @xmath72 mpc .
it has to be noted that the value of this maximum radius does not influence in practice the resulting modal radii and confidence intervals , because the probability distributions for @xmath48 usually go to zero well before this upper limit .
@xmath67 plays a significant role only if the number of misses is extremely low ( 1 or 0 ) . in the case of no hits , recipe ( 6 )
gives an incorrect result , in fact , the probability @xmath44 becomes zero for @xmath73 by definition . in this extreme occurrence
we have then to write : @xmath74 where , @xmath75 is the proper normalization factor which makes @xmath76 . the global @xmath77 for a given qso pair is finally obtained by the sum : @xmath78 in order to investigate the possible correlations of the radius @xmath48 with the los separation and with the redshift , we have applied our analysis to 7 other data samples of comparable quality , published in the literature : q1343 + 2640a , b ( crotts et al . 1994 ) , q1026 - 0045a , b ( petitjean et al . 1998 ) , q0107 - 025a , b ( d97 ) , q1517 + 2356,1517 + 2357 ( cf97 ) , q1623 + 2651a,1623 + 2653,1623 + 2651b ( cf97 ) . the qso pair q1343 + 2640a , b has been observed also by dinshaw et al .
( @xcite ) , we have chosen to use crotts et al .
( @xcite ) line list because of the larger wavelength coverage and larger number of identified lines .
the relevant information are displayed in table [ t3 ] . columns 6 and 7 report the observed number of hits and misses , respectively .
the most probable number of accidental hits for each qso pair is : 0 for 1343 + 2640a , b ; 0 for 1026 - 0045a , b ( for both @xmath79 and @xmath80 ) ; 2 for 0307 - 195a , b ( @xmath80 ) ; 6 for 0307 - 195a , b ( @xmath79 ) ; 0 for 0107 - 025a , b ; 1 for 1517 + 2356/57 ; 2 for 1623 + 2651a , b ; 2 for 1623 + 2651a/53 and 5 for 1623 + 2653,51b . in the actual computation of the cloud sizes the average value of the proper separation has been adopted . in fig .
[ pdis ] the probability distributions given by equation ( 8) for all the qso pairs presented in the paper are plotted for the two cases of spherical and disc geometry . in fig .
[ seprad : sph ] and fig [ seprad : dsk ] the values estimated for the modal radii are plotted as a function of @xmath47 .
we consider that if the _ typical _
radius @xmath81 , where @xmath82 is defined as @xmath83 only an upper limit can be set to the dimension of the absorber .
this is the case for q0307 - 195a , b ( with @xmath84 ) for which @xmath85 kpc for the spherical model and @xmath86 kpc for the disc model . the dashed line represents the most probable modal radius , @xmath4 kpc , with 95% confidence limits @xmath5 kpc ( spherical absorbers ) and @xmath6 kpc , with 95% confidence limits @xmath7 kpc ( disc ) .
these values correspond to the peak of the total probability density obtained as the product of the probability densities of the considered qso pairs .
a plot of the modal radius as a function of the average redshift of the ly@xmath0 forest of the corresponding qso pair does not show any correlation between the two , at variance with the result obtained by dinshaw et al .
( @xcite ) .
the quality of the present data on the pair q0307 - 195a , b and of the data by petitjean et al .
( @xcite ) for the pair q1026 - 0045a , b allows defining a complete sample of lines down to an equivalent width @xmath87 .
the resulting number of hits and misses , the radius and confidence interval are reported in table [ t3 ] .
the corresponding values of the global modal radius and 95 % confidence interval are @xmath88 kpc and @xmath89 kpc ( sphere ) , @xmath90 kpc and @xmath91 kpc ( disc ) .
until fairly recently it has been conjectured that the typical transverse sizes of the ly@xmath0 were of the order of a few tens of kpc .
the present analysis indicates about one order of magnitude larger dimensions at @xmath92 , implying correspondingly larger ionizations and masses . with the presently inferred sizes and the typical uv background conditions at @xmath93 ( giallongo et al . 1996 ) , optically thin absorbers of @xmath94 are expected to show a remarkably small fraction of neutral hydrogen , roughly @xmath95 . in this way the mean intergalactic density contributed by these clouds ( hu et al .
1995 ) , especially in the spherical case , is close to conflict with the baryon limit from nucleosynthesis @xmath96 ( walker et al .
this potential incongruity can however be alleviated by assuming a disc or other geometries .
-6 cm the approach described in the previous section , although customarily used to determine the size of the ly@xmath0 absorbers , is expected to give just a rough idea of the dimension of these structures , due to the simplistic hypotheses made on their geometry .
in particular , the information on the equivalent width of the lines are not taken into account , since no density distribution is assumed inside the absorbers .
besides , all the clouds are assumed to have the same radius . in the literature ( smette et al .
1992 ; smette et al .
1995 ; charlton et al . 1997 ; dinshaw et al .
1997 ) this problem has been tackled adopting a statistical technique that utilizes the information about the equivalent width of the lines .
dinshaw et al .
( 1997 ) adopted a power - law column density profile for the absorbers , @xmath97 , where @xmath98 @xmath99 for @xmath100 km s@xmath2 is the limiting column density of the sample , @xmath101 is the radius of the absorber at @xmath102 and @xmath103 , and three simple geometries with fixed radius : spheres , randomly inclined discs , and pseudo - filamentary structures ( approximated as discs with fixed inclination @xmath104 )
. the column density distribution along paired lines of sight for each of the geometric models considered is simulated by means of the monte carlo techniques outlined by smette et al .
( @xcite ) . this distribution can be converted into the corresponding equivalent width distribution adopting an approximation to the curve of growth ( chernomordik & ozernoy 1993 ) . in fig .
[ eqwdt ] the equivalent widths of the hits and misses are plotted for an enlarged sample of qso pairs formed by our previously described 8 cases plus the lensed qso he 1104 - 1805a , b observed by smette et al .
( @xcite ) .
the pairs are grouped according to their typical proper separation .
it should be noted that it is now certain that the double qso he 1104 - 1805a , b is actually a gravitational lens system , as the lensing galaxy has been detected ( courbin et al .
1998 ; remy et al . 1998 ) .
the exact values of the separation between the two light beams depend on the still badly determined lens redshift .
the values we reported in fig .
[ eqwdt ] have been estimated using fig . 6 of the paper by smette et al .
( @xcite ) , for the redshift @xmath105 , favoured by remy et al .
it is clear from the figure that the fraction of hits decreases as the proper separation increases , while , at the same time , the correlation between the equivalent widths of the absorbers in the two los becomes poorer and poorer .
the present observations can be compared with the simple model predictions of d97 ( their fig .
12 ) . in particular ,
our fig .
[ eqwdt](a ) shows reasonable agreement with the prediction of a disc model with @xmath106 that is @xmath107 kpc ; panel ( b ) agrees with the case @xmath108 for both filament and disc model , that is @xmath109 kpc ; panel ( c ) can be explained by filaments and disc with @xmath110 which implies @xmath111 kpc and , finally , panels ( d ) and ( e ) agrees with filament and disc models with @xmath112 or @xmath113 kpc . altogether a disc geometry with @xmath114 kpc seems to be favored .
it should be noted , however , that the number of degrees of freedom is not the same in the three models : once a given radius is chosen in the disc idealization the two orientation angles can vary at random , for the filament geometry one angle is fixed , and for the sphere no angular variable exists to define an orientation . in this way , it is obviously easier for the disc model to reproduce the real data that show a substantial amount of scatter due to measurement errors and possibly to a non - unique value of the cloud size .
in fact , by introducing one more degree of freedom in the spherical cloud model : i.e. allowing the radius @xmath48 to vary according to one of the two distribution : ( 1)exponential , and ( 2)power law with indices @xmath115 , it is found that a power law distribution with @xmath116 provides an equally reasonable representation of the observed equivalent width distribution ( d97 ) .
we have presented new spectra of the quasar pair q0307 - 195a , b .
the number of detected lines is almost doubled with respect to the previous study of this pair ( shaver & robertson 1983 ) .
one new metal system has been identified in object a on the basis of the doublet and one possible system is found also in object b. in order to study the shape and dimension of the ly@xmath0 absorbers , we have added to the observations of the present qso pair , data of comparable resolution taken from the literature to create a sample of 8 qso pairs .
we looked for hits and misses in the pair spectra and computed the probability distribution for a given radius @xmath48 in the case of spherical and disc geometry basing our statistical analysis on the approach outlined by fang et al .
( @xcite ) .
we modified some of their assumptions in order to deal correctly with the extreme cases of very low numbers ( 1 or 0 ) of hits or misses and avoid any spurious dependence on the separation between lines of sight .
- : : the total probability density obtained as the product of the probability densities of the considered qso pairs , gives a modal radius @xmath4 kpc , with 95% confidence limits @xmath5 kpc ( spherical absorbers ) and @xmath6 kpc , with 95% confidence limits @xmath7 kpc ( disc ) . - : : no significant correlation is detected between the modal radius and the mean proper separation of the pair , nor between the modal radius and the mean redshift of the ly@xmath0 forest region .
- : : the comparison of the distribution of the equivalent widths of the observed coincident lines with the predictions of monte carlo simulations for three simple geometrical models , spheres , discs and filaments with a power - law column density profile @xmath117 ( dinshaw et al .
1997 ) , seems to indicate a disc shape with @xmath118 kpc and @xmath119 as the most suitable .
this could however be simply the result of the different number of parameters in the three models .
if , for example , a further degree of freedom is allowed to the model based on spheres in the form of a power - law distribution of radii , a satisfactory consistency with the data is obtained .
we are grateful to j. bergeron for enlightening suggestions .
we thank the referee , dr . a. smette , for the stimulating comments .
sc acknowledges the support of the asi contract ars-96 - 176 and of the formation and evolution of galaxies network set up by the european commission under contract erb fmrx - ct96 - 086 of its tmr programme .
|
spectra of the qso pair q0307 - 195a , b have been obtained in the ly@xmath0forest ( 36603930 ) and ( 47204850 ) regions with a fwhm resolution between 0.7 and 0.5 .
46 lines have been detected in the spectrum of object a while 36 in the spectrum of object b , of them 29 and 20 were identified as ly@xmath0absorptions respectively .
the present observations have been supplemented with data of comparable quality on other 7 qso pairs available in the literature to give an enlarged sample of 217 ly@xmath0 lines with rest equivalent width @xmath1 .
the analysis of the hits ( i.e. when an absorption line appears in both qso spectra ) and misses ( i.e. when a line is seen in any of the qso spectra , but no line is seen in the other ) , carried out with an improved statistical approach , indicates that the absorbers have typically a large size km s@xmath2 mpc@xmath2 ) , @xmath3 . ] : @xmath4 kpc , with 95% confidence limits @xmath5 kpc and @xmath6 kpc , with 95% confidence limits @xmath7 kpc for the radius of idealized spherical and disc geometries , respectively .
the present data do not allow to establish any correlation of the typical inferred size with the proper separation or with the redshift of the pairs .
the correlation between the observed equivalent widths of the absorbers in the adjacent lines of sight becomes poorer and poorer with increasing proper separation .
a disc geometry with a column density profile @xmath8 , @xmath9 , is found to reasonably reproduce the data with @xmath10 kpc , but also spherical clouds with the same column density profile and a power - law distribution of radii may give a satisfactory representation of the observations .
|
the phase - ordering dynamics of systems quenched from a high - temperature disordered state into an ordered state is a problem of great relevance in the description of out - of - equilibrium pattern formation @xcite .
one well established property is the onset of dynamic scaling , where the late - time behaviour of the order - parameter correlation functions is described by scaling forms with a single time - dependent length scale @xmath14 .
thus the real - space correlation function is found to have the scaling form @xmath15 while its fourier transform , the structure factor , has the corresponding scaling form @xmath16^d g(kl(t))\ .
\label{scaling : fourier}\ ] ] conventional experimental systems such as binary alloys and binary liquids are described by a scalar order parameter .
recently , however , there has been much interest in systems with more complicated order parameters such as @xmath11-component vectors ( the @xmath0 model ) @xcite@xcite and traceless symmetric tensors ( nematic liquid crystals ) @xcite . in this paper
we restrict discussion to the @xmath0 model .
much numerical and theoretical effort has been devoted to understanding the basic properties of systems that can support singular topological defects , i.e. systems with @xmath17 .
the presence of such defects leads to a generalization of the usual porod law for the large-@xmath18 tail of the structure - factor scaling function @xmath8 , namely @xmath19 for @xmath20 @xcite .
this result is geometrical in origin , and is independent of whether or not the order parameter is conserved by the dynamics @xcite .
very recently , the cases @xmath21 , for which nonsingular topological textures occur , have been studied numerically ( for @xmath3 ) and analytically ( for @xmath5 ) .
rutenberg and bray @xcite found that the @xmath5 xy model ( @xmath22 ) exhibits a scaling violation due to the existence of _ two _ relevant length scales : the phase coherence length and phase winding length . on the other hand ,
the two - dimensional heisenberg model ( @xmath23 ) with non - conserved dynamics also violates dynamic scaling due , it appears , to the existence of as many as three separate length scales , related to individual texture size , the typical separation between textures and the typical distance between textures of opposite charge @xcite .
these texture systems are , perhaps , the most complex of the phase ordering systems . by contrast , systems without topological defects ( @xmath24 ) seem relatively straightforward .
there is good evidence for the simple scaling behavior described by ( [ scaling : real ] ) and ( [ scaling : fourier ] ) , with characteristic scale @xmath25 for nonconserved dynamics .
the energy scaling approach of bray and rutenberg @xcite shows that , provided scaling holds , @xmath25 is indeed correct for @xmath24 systems with nonconserved dynamics , and gives @xmath26 for conserved dynamics , again nicely consistent with simulation results @xcite and the renormalization group result of bray @xcite .
much less is known , however , about the form of the structure - factor tail for @xmath24 .
the recent simulations of rao and chakrabarty ( rc ) @xcite , with conserved dynamics , for the cases @xmath5 , @xmath23 and @xmath3 , @xmath27 show ` squeezed exponential ' behavior [ i.e. @xmath28 with @xmath29 . in this paper
we concentrate on systems with @xmath24 and nonconserved dynamics .
we consider the cases @xmath3 , @xmath4 and @xmath5 , @xmath6 . in each case
we confirm the expected @xmath30 growth , and find ` stretched exponential ' behavior [ i.e. @xmath28 with @xmath31 for the tail of the structure factor , with an exponent @xmath10 that appears to depend on @xmath11 and @xmath2 . in an attempt to understand the origin of this tail behaviour
, we present an analytical approach based on an approximate equation due to bray and humayun ( bh ) @xcite , which is itself based on the ` gaussian auxiliary field ' ( gaf ) method @xcite that describes rather well the form of the structure factor for nonconserved systems with singular defects ( @xmath32 ) . for those systems , this method reproduces , in particular , the generalized porod tail . for @xmath24 , the physical basis of the method is less clear .
however , the simple truncation of the equation at leading order in @xmath33 , proposed by bh in another context @xcite , leads to an exponential decay of @xmath8 , modified by a power - law prefactor for @xmath34 .
it is noteworthy that the asymptotic behaviour is nonanalytic in @xmath33 : for @xmath11 strictly infinite , the gaussian form @xmath35 is obtained .
the exponential asymptotics of the bh equation were noted in recent numerical studies by castellano and zannetti @xcite . using a ` hard - spin ' model newman et al .
@xcite studied numerically the dynamics of one - dimensional systems without defects for @xmath36 and @xmath37 .
measuring only the real - space correlations , they found that dynamic scaling is obeyed with characteristic length @xmath38 .
moreover , they found the real - space correlation function was very well fitted by a gaussian form , which is the exact result in the limit @xmath39 .
the fourier space analysis presented here , revealing stretched exponential tails , shows that the good gaussian fits achieved in real space are misleading .
our main results can be summarized as follows : ( i ) for all our models the characteristic length scale required to collapse the data for the real - space correlation function and structure factor , is @xmath40 , in agreement with theoretical predictions @xcite ; ( ii ) the asymptotic behaviour of the structure factor is well described by a stretched exponential of the form @xmath28 , where the exponent @xmath10 apparently depends on both @xmath11 and @xmath2 and seems to be different from the value obtained for the corresponding system with conserved dynamics @xcite ; ( iii ) an analytical treatment of the approximate bh equation , expected to be valid at large ( but finite ) @xmath11 , gives a simple exponential modified by a power , @xmath12 , with the _ same _ asymptotic form for conserved dynamics .
the rest of the article is organized as follows . in the next section ,
we introduce the cds model based on the time - dependent ginzburg - landau ( tdgl ) equation for a zero temperature quench , and we describe the corresponding numerical procedure employed in the simulation .
section 3 presents simulation results for a vector order parameter with @xmath27 and @xmath41 components in one and two dimensions and the one - dimensional @xmath42 model . for the @xmath3 systems , we also present data for the real - space correlation function to demonstrate the dynamic scaling behaviour .
we then discuss the procedure used to obtain the asymptotic functional form of the structure factor tail .
next we compare the data to the results of the approximate analytic theory .
finally we make some concluding comments on our results and a give a brief summary .
the dynamic evolution of a non - conserved vector order parameter ( model a ) with @xmath11 components @xmath43 , for a zero - temperature quench , is described by a purely dissipative process defined in term of the following tdgl equation : @xmath44 with @xmath45 a kinetic coefficient that we will set equal to unity , and @xmath46 the free energy functional which generates the thermodynamic force , @xmath47 = \int d^d{\bf x } \left [ \frac{1}{2 } ( \nabla \vec{\phi}({\bf x},t))^2 + v(\vec{\phi}({\bf x},t ) ) \right]\ , \label{f}\ ] ] with the potential defined as @xmath48 where @xmath49 .
the ground states , or fixed points of the dynamics , are determined by the condition @xmath50 , which defines a degenerate manifold of states connected by rotations . in the internal space of the order parameter , this manifold is the surface of an @xmath11-dimensional sphere . at late times
the order parameter is saturated in length ( i.e. lies on the ground state manifold everywhere ) .
then the dynamics is driven by the decrease of the free - energy associated with the term @xmath51 in ( [ f ] ) , through a reduction in the magnitude of the spatial gradients .
we can construct an explicit numerical scheme for the simulation based on a computationally efficient algorithm , namely the cell - dynamic system ( cds ) @xcite , which updates the order parameter according to the rule @xmath52\ , \ ] ] with @xmath53 where @xmath54 is the number of nearest neighbors , and @xmath55 and @xmath56 are parameters that we choose to be @xmath57 and @xmath58 in our simulations .
the above numerical procedure is identical to that used by toyoki @xcite , differing only in the values of the parameters @xmath55 and @xmath56 .
the cds is an euler - like algorithm and for convenience in our analysis of the results we use a unit of time equal to the update time step @xmath55 .
it should be noted ( see figures 1 and 3 ) that the scaling regime is reached very quickly in these systems without defects , and very long runs are not necessary .
the two - dimensional systems consist of a square lattice of size @xmath59 with periodic boundary conditions .
the physical quantities are calculated as averages over 20 independent distributions of initial conditions .
the one - dimensional systems have @xmath60 sites ( with periodic boundary conditions ) and we average @xmath61 independent runs .
the initial conditions for the order parameter components @xmath62 were randomly chosen from a uniform distribution with support on the interval ( -0.1,0.1 ) .
a quantity of interest that is computed during the course of the numerical simulation in the two - dimensional models is the two - point real - space correlation function @xmath63 where @xmath64 stands for the average over the set of independent initial conditions ( or ` runs ' ) . a spherical average over all possible distances @xmath65 is performed to find the isotropic real - space correlation @xmath66
. the other function of interest , calculated for all the models , is the structure factor , @xmath67 we also make a spherical average over all possible values of @xmath68 with given @xmath69 . in the calculation of these quantities at each time , the data are ` hardened ' by replacing the order parameter at each point by a unit vector in the same direction ( the fixed point of the cds iteration being a vector of unit length ) .
this procedure accelerates the entry into the dynamic scaling regime , and helps us to elucidate the proper nature of the asymptotic tail in the structure factor .
dynamic scaling is observed for all the models studied .
the scaling regime is reached at quite early times , in agreement with previous studies .
we show that dynamic scaling holds in the two - dimensional systems ( @xmath4 ) , using the characteristic length @xmath7 deduced from theoretical considerations @xcite .
this agrees with earlier simulations of bray and humayun using ` hard - spin ' dynamics @xcite .
figure 1(a ) presents a plot for @xmath3 , @xmath27 of the correlation function ( [ scaling : real ] ) as a function of distance @xmath70 for several times , while in figure 1(b ) we show the collapsed dynamic scaling function when the analysis is made using the scaling variable @xmath71 . as can be seen from the figure ,
the scaling function @xmath72 is a monotonically decreasing function with the generic featureless shape that is characteristic of non - conserved @xmath0 models .
it is of some interest to investigate the small-@xmath73 behavior of the real - space scaling function @xmath72 . in systems with @xmath17 , the existence of singular topological defects leads to a non - analytic term of the form @xmath74 ( with an additional @xmath75 factor for even @xmath11 ) , which leads to the @xmath76 porod tail in fourier space @xcite . in the present case , where @xmath77 , we expect no such short - distance singularities . therefore , we consider an expansion of the form @xmath78 . in table 1
we present the parameters @xmath79 and @xmath80 determined from the simulations in the range @xmath81 .
the ratio @xmath82 should be a universal number for given @xmath11 and @xmath2 .
it will be seen from table 1 that this ratio has the value @xmath83 for @xmath27 , different from the value @xmath84 obtained for a gaussian function , which is the exact result for the limit @xmath39 . for @xmath41 ,
table 1 gives @xmath85 , already consistent with the large-@xmath11 result .
however , in the absence of any short - distance singularity , the small-@xmath73 behavior provides no useful information on the nature of the tail in the structure factor .
consequently , it is more convenient to investigate directly the simulation results of the structure factor and extract from them the asymptotic behaviour .
we shall see that the behavior in fourier space is clearly non - gaussian , even for @xmath41 .
as expected , given the absence of topological defects , the results indicate ( fig .
2 ) that the decay of the structure factor is clearly faster than a power law , in contrast to the interpretation of his own similar results by toyoki @xcite . in order to demonstrate that the tail is well described by the stretched exponential form @xmath86 where @xmath87 is the scaling variable in momentum space , we attempt to find the corresponding power @xmath10 in the exponential by plotting @xmath88 versus @xmath89 and adjusting the value of @xmath10 until the best linear behavior is obtained in the regime @xmath90 . during the fitting procedure
the other two parameters of the fit , @xmath91 and @xmath92 , are readily determined .
the criteria used for the optimum fitting is based on the pearson correlation coefficient ( pcc ) , which measure the strength of the linear relation among two variables and varies between -1 ( perfect negative linear relation ) and @xmath93 ( perfect positive linear relationship ) .
we proceed as follows : first , we choose an exponent @xmath10 and then perform linear regression ; next we change @xmath10 until the pcc reaches its maximum value .
the regression coefficients are calculated using the values of the scaling structure function at the last two times in the simulation .
the optimum values for system with @xmath27 components are @xmath94 with a pearson coefficient of @xmath95 .
the other two parameters are @xmath96 and @xmath97 this result is presented in the fig .
we turn now to the description of the case @xmath41 , following a similar analysis to the @xmath27 model .
3(a ) shows the correlation function as a function of distance for different times . in this model
the collapse is also achieved using the characteristic length @xmath40 , as can be observed in fig.3(b ) .
therefore , both models are consistent with dynamical exponent @xmath98 . the corresponding scaling plot for the structure factor is shown in fig .
a more important effect is observed in the structure factor tail , because in this case it has also a stretched exponential but with an apparently larger exponent . following an analysis similar to that used for @xmath27 , we find that the value of the best fit value of the exponent is @xmath99 , and the corresponding pcc in the regression is @xmath100 .
the other two parameters are @xmath101 and @xmath102 . in fig .
4(b ) we plot @xmath88 against @xmath89 and the linear behaviour is clearly seen .
comparison between the real - space correlation functions of the @xmath27 and @xmath41 models shows that the scaling functions are very similar ; the main difference is that the scaling function decreases slightly more slowly for @xmath41 than @xmath27 .
this is reflected in the parameters of the fitting function for the small-@xmath73 range : the amplitudes @xmath79 and @xmath80 tend to decrease as @xmath11 increases ( table 1 ) .
we turn to the discussion of the simulation results for one - dimensional systems .
we shall describe the relevant behavior in fourier space .
real - space data has been presented in @xcite .
our results show that in one - dimensional systems the asymptotic behavior of the structure factor also has a stretched exponential form , but the fitted exponents @xmath10 are larger than those of the corresponding two - dimensional models , and close to unity for @xmath27 and 5 . in fig .
5(a ) , we present the simulation results for the scaling function @xmath8 of the structure factor for the one - dimensional heisenberg model ( @xmath23 ) .
the continuous curve is the result of the analytical approach described in section iv .
the analysis of the asymptotic behaviour gives an exponent @xmath103 for the stretched exponential .
5(b ) shows the curve of @xmath104 $ ] versus @xmath89 , where the linear behaviour is clearly observed .
similarly , we present the corresponding plots for the @xmath27 model in fig . 6 , where the measured exponent is now @xmath105 , while for @xmath41 we obtain @xmath106 as is shown in fig .
the values of @xmath10 for the last two models are so close that in practice it is difficult to distinguish between them .
they are also close to the value unity obtained from the approximate large-@xmath11 equation discussed in the following section .
it is clear from the results for the one - dimensional models that the scaling function in real space is not a gaussian function , despite the good real - space fits to this form obtained in @xcite .
moreover , the ( effective ) exponents @xmath10 for the @xmath4 models are bigger than for the corresponding models in two dimensions .
therefore , we have evidence that the exponent @xmath10 increases with @xmath11 , while it seems to decrease with @xmath2 .
note that the gaussian result obtained for @xmath107 corresponds to @xmath108 , so the results presented here for the structure - factor tail are actually quite far from that limit . the analytical treatment presented in the next section
gives some indication of why this might be expected .
in particular , it suggests that the structure factor is dominated by a simple exponential for @xmath109 at fixed large @xmath11 , while the familiar gaussian form is recovered as @xmath39 at fixed @xmath18 .
we conclude this section by discussing briefly some alternative fitting forms for the structure factor tail .
first , however , we note that the stretched exponential form ( [ se ] ) describes the tail well over at least 10 decades of @xmath110 in all cases . of course , this represents a much smaller dynamic range ( 1 to @xmath111 decades ) in the scaling variable @xmath87 .
motivated by the analytical result [ equation ( [ g ] ) below ] of the approximate large-@xmath11 theory , other fitting forms were tried for @xmath3 ( the agreement with the large-@xmath11 theory already being good for @xmath112 and 5 in @xmath5 )
. a direct fit of ( [ g ] ) does not work well for @xmath3 .
allowing for a general power - law prefactor , @xmath113 , gives a reasonable fit , but with very large values for @xmath73 5.6 for @xmath27 , and 6.7 for @xmath41 .
fixing @xmath114 , but allowing for a general stretched exponent @xmath10 again gives a reasonable fit ( with @xmath115 for @xmath27 and 0.70 for @xmath41 ) , but over a significantly reduced range of @xmath18 .
for these reasons we prefer the unmodified stretched exponential ( [ se ] ) as giving the simplest and most convincing description of the large-@xmath18 data , at least for these small values of @xmath11 in @xmath3 .
of course , it is quite possible that the form ( [ g ] ) will fit the data well at larger values of @xmath11 .
in an attempt to gain some analytical insight into the structure factor asymptotics , we start from an approximate equation of motion for the pair correlation function derived using the gaussian auxiliary field approach pioneered by mazenko @xcite .
we then make , for reasons that will become clear , the further simplification of retaining only the leading nonlinearity as @xmath39 .
the resulting equation is then finally used to extract the asymptotics of @xmath8 .
the gaf method for vector fields has been discussed in some detail elsewhere .
we refer the reader to the original papers @xcite and a recent review @xcite for a full exposition .
the essence of the method is a mapping from the original field variable @xmath116 to an ` auxiliary field ' @xmath117 .
the function @xmath118 satisfies the equation @xmath119 , where @xmath120 is the potential in the ginzburg - landau function ( 4 ) . with the boundary conditions @xmath121 , and @xmath122 for @xmath123 , this equation for @xmath118 represents the equilibrium profile function for a spherically symmetric topological defect , with @xmath124 representing distance from the defect .
the ( uncontrolled ) approximation that @xmath117 is a gaussian field , and the imposition of the scaling form ( 1 ) , leads eventually to the self - consistent equation @xcite @xmath125 for the scaling function @xmath72 , where primes indicate derivatives . in ( [ sc1 ] )
@xmath126 is the normalized correlator of the auxiliary field , @xmath127^{1/2}$ ] , where ` 1 ' and ` 2 ' represent the space - time points @xmath128 and @xmath129 , and the function @xmath130 is given by @xmath131 ^ 2\ , f\left(\frac{1}{2},\frac{1}{2};\frac{n+2}{2};\gamma^2\right)\ , \label{bpt}\ ] ] where @xmath132 is the beta function , and @xmath133 the hypergeometric function .
the constant @xmath134 in ( [ sc1 ] ) has to be adjusted so that @xmath72 vanishes sufficiently fast at infinity @xcite .
as should be clear from the above discussion , ( [ sc1 ] ) only really makes sense for @xmath17 , based as it is on the presence of singular topological objects whose positions are defined by the zeros of the field @xmath116 or , equivalently , by the zeros of @xmath117 .
indeed , the function @xmath130 has inbuilt structure that generates the porod tail associated with such defects .
specifically , in the short - distance limit , where @xmath135 , the hypergeometric function in ( [ bpt ] ) has a singular contribution of order @xmath136 ( with a logarithmic correction for even @xmath11 ) .
since @xmath137 for small scaling variable @xmath73 , this singular term is of order @xmath138 ( again , with a logarithm for even @xmath11 ) , leading to the power - law tail @xmath19 in fourier space . within the gaf approach , this tail is obtained for _ all _ @xmath11 and @xmath2 . for @xmath24 , however , neither singular topological objects nor nonsingular topological textures exist , so the gaf result is qualitatively incorrect . indeed
, this is to be expected since the gaf approach is specifically designed to build in the defect structure .
so what should one do when there are no defects ?
we have seen that the usual gaf approach always give a porod tail , for any @xmath11 and @xmath2 : this is unphysical for @xmath24 , since the tail is a consequence of the presence of topological defects .
one way around this impasse is to artificially approximate the full gaf equation ( [ sc1 ] ) by the form valid for @xmath39 . in this limit @xmath139 , and ( [ sc1 ] )
becomes , correct to @xmath140 , @xmath141 this step , admittedly ad - hoc , has the desired effect of eliminating the unwanted ( for @xmath24 ) short - distance singularity in @xmath72 .
( [ sc2 ] ) is the nonconserved version of the equation introduced by bh to study the crossover from multiscaling to simple scaling in the asymptotic dynamics of a _ conserved _ vector field at large but fixed @xmath11 @xcite .
both conserved and nonconserved versions have recently been studied numerically @xcite . to extract analytically the large-@xmath18 behavior , we perform a ( @xmath2-dimensional ) fourier transform of ( [ sc2 ] ) .
the resulting equation for @xmath142 is @xmath143 where @xmath144 if we assume an asymptotic form @xmath145 , with @xmath146 , then ( [ sc3 ] ) gives @xmath147 in the appendix , we show that consistency with ( [ largeq ] ) requires @xmath148 and @xmath149 , i.e. @xmath150 in real - space this implies that the function @xmath151 has simple poles in the complex @xmath54 plane at @xmath152 .
the value of @xmath92 is not determined by this argument ; instead one can derive ( see appendix ) the relationship @xmath153 between @xmath92 and the prefactor @xmath91 in the asymptotic form ( [ g ] ) .
the existence of these simple poles in real space also follows directly from the real - space equation ( [ sc2 ] ) . if one assumes a singularity of the form @xmath154 , with @xmath155 , then balancing the dominant terms @xmath156 and @xmath157 gives immediately @xmath158 , i.e. a simple pole .
the position @xmath159 is not determined , but the residue @xmath160 of the pole is given by @xmath161 , where the two values correspond to the poles @xmath162 . using this result for @xmath160
, one can readily recover ( [ a ] ) by contour methods , e.g. for @xmath5 one has @xmath163 where the second line , equivalent to ( [ a ] ) for @xmath5 , was obtained by closing the contour in the upper half - plane .
the approach outlined above gives the relation ( [ a ] ) between @xmath91 and @xmath92 , but does not determine @xmath92 explicitly .
we now give a heuristic argument that @xmath13 for large @xmath11 .
first we make an observation concerning the value of @xmath134 .
equation ( [ sc3 ] ) with @xmath164 gives @xmath165^{-1}\ .\end{aligned}\ ] ] in particular , @xmath166 for @xmath107 .
for @xmath107 , therefore , ( [ sc3 ] ) becomes @xmath167 , with solution @xmath168 [ the prefactor being fixed by the condition @xmath169 .
for @xmath11 large but finite , on the other hand , we have seen that the asymptotic form is @xmath170 .
the crossover between these two forms presumably occurs at some @xmath171 , with @xmath35 for @xmath172 , and @xmath170 for @xmath173 .
matching these two forms at @xmath174 gives @xmath175 .
next we evaluate @xmath176 in the region @xmath177 . here
@xmath178 , so @xmath179 , giving @xmath180 .
however , this decays more slowly with @xmath18 than the other terms in ( [ sc3 ] ) , which fall off as @xmath181 .
so the term involving @xmath176 ( evaluated for @xmath177 ) becomes comparable with the other terms when @xmath182 , i.e. when @xmath183 .
this suggests @xmath184 , and therefore @xmath13 . the numerical data of castellano and
zannetti @xcite certainly show that @xmath92 increases extremely slowly with @xmath11 . to compare this approximate theory with our simulation data , we have solved ( [ sc1 ] ) numerically for @xmath5 , @xmath6 , using the procedure described in @xcite .
the fourier transform was then taken numerically , and the ` best fit by eye ' to the structure - factor data was obtained by adjusting the timescale in the theoretical curves , giving the results shown by the continuous curves in figures 5(a ) , 6(a ) and 7(a ) .
the corresponding log - linear plots , which reveal the large-@xmath18 behaviour more clearly , are shown in figures 8(a)-(c ) .
as might be expected , equation ( [ sc1 ] ) [ or its fourier transform ( [ sc3 ] ) ] does not describe the data quantitatively over the whole range of @xmath87 , but it does give a qualitatively correct description .
there is an early parabolic region , corresponding to a gaussian form for @xmath8 , which then gives way to a slower decay that , at least for @xmath27 and 5 , is consistent with the simple exponential form predicted by ( [ sc3 ] ) but with a different coefficient @xmath92 in the exponent .
given that the theory is , at best , a large-@xmath11 theory we regard these results as encouraging .
the @xmath23 data , however , and the @xmath3 data , do not seem to fit a simple exponential , at least for the range of @xmath18 that we have been able to explore .
( this is of course implicit in the values @xmath185 obtained for these systems from figures 2,4 and 5 . )
it may well be that considerably larger values of @xmath11 are needed in @xmath3 than in @xmath5 for the large-@xmath11 asymptotics to become apparent .
the above derivation of an exponential tail was specific to nonconserved fields .
what can we say for conserved fields ?
the fundamental equation of motion for this case is obtained from the tdgl equation ( [ tdgl ] ) by the replacement @xmath186 . applying the gaf method to this equation , imposing the scaling form ( [ scaling : real ] ) [ but with @xmath187 for conserved fields ] , and taking the fourier transform , leads to @xcite @xmath188 instead of ( [ sc3 ] ) .
[ the definition ( [ b ] ) of @xmath176 differs by a constant from that used in @xcite , where @xmath134 was written as @xmath189 , @xmath190 being the position of the maximum of @xmath8 for large @xmath11 . ] assuming the asymptotic form @xmath191 for @xmath109 , ( [ cons ] ) gives ( [ largeq ] ) once more , provided @xmath192 .
then our previous arguments apply , and the asymptotic form ( [ g ] ) , with @xmath91 and @xmath92 related by ( [ a ] ) , are recovered .
this approach therefore predicts that the structure factors for conserved and nonconserved systems will have the _ same _ asymptotic forms , at least within the context of the bh truncation .
the same conclusion was drawn from recent numerical solutions of the bh equation @xcite .
in summary , we have studied the dynamics of phase ordering for models without topological defects in one and two dimensions .
we find that scaling is achieved with the growth law @xmath40 .
the tail in the structure factor is well fitted by a stretched exponential form .
for the two - dimensional systems , table 1 summarizes the relevant parameters describing the fits in real and fourier space .
in contrast to systems with singular defects ( @xmath32 ) , where the generalized porod form @xmath19 for the structure factor tail is a consequence of the defect structure , and is independent of the presence or absence of conservation laws , in systems without defects the functional form does , apparently , differ for conserved and nonconserved systems .
we have shown , for example , that for the particular case of the @xmath27 model in two dimensions the tail is well described , over the range of @xmath18 accessible to us , by a stretched exponential with exponent @xmath193 , differing from the result for the same model with conservation studied by rc @xcite , who found @xmath194 . within the ` toy ' equation of bray and humayun @xcite , however , we have shown that the true asymptotics are the _ same _ for conserved and nonconserved dynamics .
of course , the bh equation is at best a large-@xmath11 theory , and the numerical results for nonconserved and conserved dynamics may converge as @xmath11 is increased .
a related question is whether the exponents @xmath10 measured here and in @xcite are genuine asymptotic exponents , or effective exponents whose values will change as the range of @xmath18 over which the fit is made is moved to larger @xmath18 .
more extensive simulations may cast some light on this issue .
the ` universal ' ( independent of conservation laws ) porod tail behavior obtained for @xmath17 is geometrical in origin , being a consequence of the field structure induced by singular topological defects @xcite .
as yet , however , we have no corresponding physical picture in the absence of topological defects .
it is interesting that , within the simple model of equation ( [ sc1 ] ) , the exponent @xmath10 jumps discontinuously from @xmath108 at @xmath107 to @xmath148 for @xmath11 large but finite .
more precisely , one can say that @xmath108 corresponds to the limit @xmath39 at fixed , large @xmath18 , while @xmath195 corresponds to @xmath109 at fixed , large @xmath11 .
we have argued that the crossover between these limiting forms for fixed , large @xmath11 occurs at @xmath183 .
this change of behavior depending on the order of the limits is reminiscent of the result obtained from the conserved version of ( [ sc1 ] ) , where a novel ` multiscaling ' behavior is obtained for @xmath39 at fixed , large @xmath196 @xcite , while simple scaling is recovered for @xmath197 at fixed , large @xmath11 @xcite . for the nonconserved case , one always has simple scaling . for both conserved and nonconserved fields , however , the asymptotics of @xmath8 are sensitive to whether @xmath11 is large or truly infinite .
this rules out , for example , exploring the asymptotics by expanding around the large-@xmath11 solution in powers of @xmath33 .
we thank sanjay puri for stimulating discussions during the early stages of this work .
f. rojas thanks conacyt ( mexico ) for financial support .
this work was supported by epsrc ( uk ) grant gr / j24782 .
in this appendix we use ( [ largeq ] ) to derive the asymptotic form ( [ g ] ) for @xmath8 . from the definition ( [ b ] )
we have @xmath198 where @xmath199 .
inserting the asymptotic form @xmath200 gives @xmath201\ , \ ] ] where latexmath:[\ ] ] but ( [ largeq ] ) implies , asymptotically , @xmath232 comparing ( [ b1 ] ) and ( [ b2 ] ) gives @xmath149 and eq .
( [ a ] ) for the amplitude @xmath91 .
a. p. y. wong , o. wiltzius and b. yurke , phys .
lett . * 68 * , 3583 ( 1992 ) ; n. mason , a. n. pargellis and b. yurke , phys .
lett . * 70 * , 190 ( 1993 ) ; r. e. blundell and a. j. bray , phys .
a * 46 * , r6154 ( 1992 ) .
fig . 1 . ( a ) real - space correlation function @xmath233 for a 2d system with @xmath27 components as a function of distance @xmath70 for several times @xmath196
the data were obtained from lattices of size @xmath234 , averaged over 20 different initial conditions .
( b ) demonstration of dynamic scaling , @xmath235 , with @xmath40 .
fig . 2 .
( a ) scaling structure factor @xmath236^{-2}\,s(k , t)$ ] as a function of scaled momentum @xmath87 for a 2d system with @xmath27 components for lattices of size 256@xmath237256 ( averaged over 20 different initial conditions ) .
( b ) demonstration of the stretched exponential behaviour , plotting @xmath238 against @xmath239 with @xmath94 .
the line is included as a guide to the eye .
( a ) scaling structure factor @xmath236^{-1}\,s(k , t)$ ] as a function of scaled momentum @xmath87 for a 1d heisenberg model [ @xmath42 model ] for lattices of size 16384 ( averaged over 100 different initial conditions ) .
continuous curve : result of the approximate analytical treatment described in the text .
( b ) demonstration of the stretched exponential behaviour , plotting @xmath238 against @xmath239 with @xmath103 .
the line is a guide to the eye .
log - linear plot of the scaled structure factor against scaled momentum for the 1d systems : ( a ) @xmath23 ( b ) @xmath27 ( c ) @xmath41 . in each case
the continuous curve is the result of the approximate theory of section iv .
|
using a cell dynamic system ( cds ) simulation scheme , we investigate the phase - ordering dynamics of non - conserved @xmath0 models without topological defects , i.e. for @xmath1 where @xmath2 is the spatial dimensionality .
in particular , we consider zero - temperature quenches for @xmath3 , @xmath4 , and for @xmath5 , @xmath6 .
we find , in agreement with previous simulations using fixed - length spins , that dynamical scaling is obtained , with characteristic length @xmath7 .
we show that the asymptotic behaviour of the structure - factor scaling function @xmath8 is well fitted by the stretched exponential form @xmath9 , with an exponent @xmath10 that appears to depend on both @xmath11 and @xmath2 .
an analytical treatment of an approximate large-@xmath11 equation for the pair correlation function yields @xmath12 , with @xmath13 for large @xmath11 , in agreement with recent simulations of the same equation .
|
we develop a bundle technique to solve nonconvex variational problems arising in contact mechanics and in other applied fields .
we are specifically interested in the delamination of composite structures with an adhesive bonding under destructive loading , a failure mode which is studied in the material sciences .
when the properties of the interlayer adhesive between the bonded bodies are given in the form of a physical law relating the normal component of the stress vector to the relative displacement between the upper and lower boundaries at the crack tip , the challenge is to compute the displacement and stress fields in order to assess the reactive destructive forces along the contact boundary , as the latter are difficult to measure in situ .
this leads to minimization of an energy functional , where a specific form of nonsmoothness arises in the boundary integral at the contact boundary .
after discretization via piecewise linear finite elements using the trapezoidal quadrature rule , this leads to a finite - dimensional nonsmooth optimization problem of the form @xmath2 where @xmath3 is locally lipschitz and neither smooth nor convex . depending on the nature of the frictional forces , the criterion
@xmath3 may be upper-@xmath0 or lower-@xmath0 , see e.g. figure [ dellaw ] .
as these two classes of nonsmooth functions behave substantially differently when minimized , we are forced to expand on existing bundle strategies and develop an algorithm general enough to encompass both types of nonsmoothness .
we prove its convergence to a critical point in the sense of subsequences , and show that it provides satisfactory numerical results in a simulation of the double cantilever beam test @xcite , one of the most popular destructive tests to qualify structural adhesive joints .
the difficulty in nonconvex bundling is to provide a suitable cutting plane oracle which replaces the no longer available convex tangent plane .
one of the oldest oracles , discussed already in mifflin @xcite , and used in the bundle codes of lemarchal and sagastizbal @xcite , or the bt - codes of zowe @xcite , uses the method of _ downshifted tangents_. while these authors use linesearch with armijo and wolfe type conditions , which allows only weak convergence certificates in the sense that _ some _ accumulation point of the sequence of serious iterates is critical , we favor proximity control in tandem with a suitable backtracking strategy .
this leads to stronger convergence certificates , where _ every _ accumulation point of the sequence of serious iterates is critical .
for instance , in @xcite a strong certificate for downshifted tangents with proximity control was proved within the class of lower-@xmath0 functions , but its validity for upper-@xmath0 criteria remained open .
an oracle for upper-@xmath0 functions with a rigorous convergence theory can be based on the _ model approach _ of @xcite , but the latter is not compatible with the downshift oracle .
to have two strings to one bow is unsatisfactory , as one could hardly expect practitioners to select their strategy according to such a distinction , which might not be easy to make in practice . in this work
we will resolve this impasse and present a cutting plane oracle based on downshifted tangents , which leads to a bundle method with strong convergence certificate for both types of nonsmoothness . in its principal components
our method agrees with existing strategies for downshifted tangents , like @xcite , and could therefore be considered as a justification of this technique for a wide class of applications .
differences with existing methods occur in the management of the proximity control parameter , which in our approach has to respect certain rules to assure convergence to a critical point , without impeding good practical performance .
the structure of the paper is as follows .
c1 ] gives some preparatory information on lower- and upper-@xmath0 functions .
section [ algorithm ] presents the algorithm and comments on its ingredients .
theoretical tools needed to prove convergence are presented and employed in sections [ modelconcept ] and [ oracle ] .
section [ main ] gives the main convergence result , while section [ practical ] discusses practical aspects of the algorithm .
in section [ delamination ] , we discuss the delamination problem , which we solve numerically using our bundle algorithm . numerical results for contact problem with adhesion based on the bundle - newton method of l. lukan and j. vlek @xcite can be found e.g. in the book of haslinger et al .
@xcite , in @xcite , and in the more recent @xcite .
mathematical analysis and numerical results for quasistatic delamination problems can be found in @xcite .
following spingarn @xcite , a locally lipschitz function @xmath4 is lower-@xmath0 at @xmath5 , if there exists a compact hausdorff space @xmath6 , a neighborhood @xmath7 of @xmath5 , and a mapping @xmath8 such that both @xmath9 and @xmath10 are jointly continuous and @xmath11 is satisfied for @xmath12 .
the function @xmath3 is upper-@xmath0 at @xmath5 if @xmath13 is lower-@xmath0 at @xmath5 . in a minimization problem ( [ program ] )
, we expect lower- and upper-@xmath0 functions to behave completely differently . minimizing a lower-@xmath0 function ought to lead to real difficulties , as on descending we move _ into _ the zone of nonsmoothness , which for lower-@xmath0 goes downward .
in contrast , upper-@xmath0 functions are generally expected to be well - behaved , as intuitively on descending we move _ away _ from the nonsmoothness , which here goes upward .
the present application shows that this argument is too simplistic .
minimization of upper-@xmath0 functions leads to real difficulties , which we explain subsequently . in delamination for composite materials we encounter objective functions of the form @xmath14 where @xmath15 gathers the smooth part , while the integral term , due to the minimum , is responsible for the nonsmoothness .
let us first prove ( [ struct ] ) .
for @xmath16 and fixed @xmath17 the function @xmath18 is measurable , and since @xmath19 , it is also integrable .
hence @xmath20 is well - defined , and clearly @xmath21 , so we have @xmath22 . to prove the reverse estimate , fix @xmath17 and consider the closed - valued multifunction
@xmath23 \to 2^i$ ] defined by @xmath24 . since the @xmath25 are measurable and @xmath26 is finite , @xmath27 is a measurable multifunction .
choose a measurable selection @xmath28 , that is , @xmath16 satisfying @xmath29 for every @xmath30 $ ] . then clearly @xmath31 .
this proves ( [ struct ] ) .
let us now show that @xmath3 is upper-@xmath0 .
we consider @xmath32 in view of @xcite @xmath33 is upper-@xmath0 and its clarke subdifferential @xmath34 is strictly supermonotone uniformly over @xmath30 $ ] . by theorem 2 in @xcite , @xmath33 is approximately concave uniformly over @xmath30 $ ] .
integration with respect to @xmath30 $ ] then yields an approximately concave function with respect to @xmath35 , which by the equivalences in @xcite and @xcite is upper-@xmath0 .
note that the minimum ( [ struct ] ) is semi - infinite even though @xmath26 is finite .
minimization of ( [ structure ] ) can not be converted into a nlp , as would be possible in the min - max case .
the representation ( [ struct ] ) highlights the difficulty in minimizing ( [ structure ] ) .
minimizing a minimum has a disjunctive character , and due to the large size of @xmath36 this could lead to a combinatorial situation with intrinsic difficulty .
the model of a nonsmooth function was introduced in @xcite and is a key element in understanding the bundle concept .
[ compare @xcite ] a function @xmath37 is called a model of the locally lipschitz function @xmath38 on the set @xmath39 if the following axioms are satisfied : 1 . for every @xmath40 the function @xmath41 is convex , @xmath42 and @xmath43 .
2 . for every @xmath40 and every @xmath44 there exists @xmath45 such that @xmath46 for every @xmath47 .
the function @xmath48 is jointly upper semicontinuous , i.e. , @xmath49 on @xmath50 implies @xmath51 .
@xmath52 we recall that every locally lipschitz function @xmath3 has the so - called _ standard model _
@xmath53 where @xmath54 is the clarke directional derivative of @xmath3 at @xmath35 in direction @xmath55 .
the same function @xmath3 may in general have several models @xmath48 , and following @xcite , the standard @xmath56 is the smallest one .
every model @xmath48 gives rise to a bundle strategy .
the question is then whether this bundle strategy is successful .
this depends on the following property of @xmath48 .
a model @xmath48 of @xmath3 on @xmath57 is said to be strict at @xmath58 if axiom @xmath59 is replaced by the stronger 1 . for every @xmath60 there exists @xmath45 such that @xmath46 for all @xmath61 .
we say that @xmath48 is a strict model on @xmath57 , if it is strict at every @xmath58 .
@xmath52 we may write axiom @xmath59 in the form @xmath62 for @xmath63 , and @xmath64 as @xmath65 for @xmath66 . except for the fact that these concepts are one - sided , this is precisely the difference between differentiability and strict differentiability . hence the nomenclature .
[ compare @xcite ] suppose @xmath3 is upper-@xmath0 .
then its standard model @xmath56 is strict , and hence every model @xmath48 of @xmath3 is strict .
@xmath52 for convex @xmath3
the standard model @xmath56 is in general not strict , but @xmath3 may be used as its own model @xmath67 . for nonconvex @xmath3 , a wide range of applications
is covered by composite functions @xmath68 with @xmath69 convex and @xmath9 differentiable . here
the so - called natural model @xmath70 can be used , because it is strict as soon as @xmath9 is class @xmath0 .
this includes lower-@xmath71 functions in the sense of @xcite , lower-@xmath72 functions in the sense of @xcite , or amenable functions in the sense of @xcite , which allow representations of the form @xmath73 with @xmath9 of class @xmath74 .
we conclude with the remark that lower-@xmath0 functions also admit strict models , even though in that case the construction is more delicate . the strict model in that case
can not be exploited algorithmically , and for lower-@xmath0 functions we prefer the oracle concept , which will be discussed in section [ oracle ] .
in this section we briefly explain the main features of the algorithm .
this concerns building the working model , computing the solution of the tangent program , checking acceptance , updating the working model after null steps , and the management of the proximity control parameter . at the current serious iterate
@xmath35 the inner loop of the algorithm at counter @xmath75 computes an approximation @xmath76 of @xmath3 in a neighborhood of @xmath35 , called a first - order working model .
the working model is a polyhedral convex function of the form @xmath77 where @xmath78 is a finite set of affine functions @xmath79 satisfying @xmath80 , referred to as _
planes_. the set @xmath78 is updated during the inner loop @xmath75 . at each step
@xmath75 the following rules have to be respected when updating @xmath81 into @xmath82 : 1 .
one or several cutting planes at the null step @xmath83 , generated by an abstract cutting plane oracle , are added to @xmath82 .
the so - called aggregate plane @xmath84 , which consists of convex combinations of elements of @xmath81 , is added to @xmath82 .
some older planes in @xmath81 , which become obsolete through the addition of the aggregate plane , are discarded and not kept in @xmath82 .
4 . every @xmath78 contains at least one so - called exactness plane @xmath85 , where exactness plane means @xmath86 , @xmath87 .
this assures @xmath88 , hence the name .
we have to make sure that each working model @xmath89 satisfies @xmath90 .
once the first - order working model @xmath76 has been built , the second - order working model @xmath91 is of the form @xmath92 where @xmath93 is a possibly indefinite symmetric matrix , depending only on the current serious iterate @xmath35 , and fixed during the inner loop @xmath75 .
the second - order term includes curvature information on @xmath3 , if available .
once the second - order working model ( [ second ] ) is formed and the proximity control parameter @xmath94 is updated , we solve the tangent program @xmath95 here the proximity control parameter @xmath96 satisfies @xmath97 , which assures that ( [ tangent ] ) is strictly convex and has a unique solution , @xmath83 , called the _ trial step_. the trial step is a candidate to become the new serious iterate @xmath98 . in order to decide whether @xmath83 is acceptable
, we compute the test @xmath99 where @xmath100 is some fixed parameter . if @xmath101 , then @xmath102 is accepted and called a _
serious step_. in this case the inner loop ends successfully . on the other hand , if @xmath103 , then @xmath83 is rejected and called a _
null step_. in this case the inner loop @xmath75 continues .
this means we will update working model @xmath104 , adjust the proximity control parameter @xmath105 , and solve ( [ tangent ] ) again .
note that the test ( [ rho ] ) corresponds to the usual armijo descent condition used in linesearches , or to the standard acceptance test in trust region methods .
suppose the trial step @xmath83 fails the acceptance test ( [ rho ] ) and is declared a null step .
then the inner loop has to continue , and we have to improve the working model at the next sweep in order to perform better .
since the second - order part of the working model @xmath106 remains invariant , we will update the first - order part only .
concerning rule @xmath107 , by the necessary optimality condition for ( [ tangent ] ) , there exists a multiplier @xmath108 such that @xmath109 or what is the same , @xmath110 since @xmath76 is by construction a maximum of affine planes , we use the standard description of the convex subdifferential of a max - function .
writing @xmath111 for @xmath112 , we find non - negative multipliers @xmath113 summing up to 1 such that @xmath114 and in addition , @xmath115 for all @xmath116 with @xmath117 .
we say that those planes which are active at @xmath83 are _ called by the aggregate plane_. in the above rule @xmath118 we allow those to be removed from @xmath81 .
we now define the aggregate plane as : @xmath119 note that by construction the aggregate plane @xmath120 at null step @xmath83 satisfies @xmath121 .
this construction is standard and follows the original idea in kiwiel @xcite .
it assures in particular that @xmath122 .
the crucial improvement in the first - order working model is in adding a cutting plane which cuts away the unsuccessful trial step @xmath83 according to rule @xmath123 .
we shall denote the cutting plane as @xmath124 .
the only requirement for the time being is that @xmath125 , as this assures @xmath126 .
since we also maintain at least one exactness plane of the form @xmath127 with @xmath128 , we assure @xmath129 .
later we will also have to check the validity of @xmath130 .
it is possible to integrate so - called _ anticipated cutting planes _ in the new working model @xmath82 .
here anticipated designates all planes which are not based on the rules exactness , aggregation , cutting planes . naturally , adding such planes can not be allowed in an arbitrary way ,
because axioms @xmath131 have to be respected .
it may be beneficial to choose a new exactness plane @xmath132 after each null step @xmath133 , namely the one which satisfies @xmath134 .
if @xmath35 is a point of differentiability of @xmath3 , then all these exactness planes are identical anyway , so no extra work occurs . on the other hand , computing @xmath135 such that @xmath136 is usually cheap .
consider for instance eigenvalue optimization , where @xmath137 , @xmath17 , @xmath138 , and @xmath139 is the maximum eigenvalue function of @xmath140 .
then @xmath141 , where @xmath142 , @xmath143 , and where @xmath144 is a @xmath145 matrix whose columns form an orthogonal basis of the maximum eigenspace of @xmath146 of dimension @xmath147 @xcite .
then @xmath148 attains @xmath149 , hence @xmath150 attains @xmath151 . since usually @xmath152 ,
the computation of @xmath69 is cheap .
the central novelty of the bundle methods developed in @xcite is the discovery that in the absence of convexity the proximity control parameter @xmath153 has to follow certain basic rules to assure convergence of the sequence @xmath154 of serious iterates .
this is in contrast with convex bundle methods , where @xmath153 could in principle be frozen once and for all .
more precisely , suppose @xmath76 has failed and produced only a null step @xmath83 .
having built the new model @xmath155 , we compute the secondary test @xmath156 where @xmath157 is fixed .
our decision is @xmath158 the rationale of ( [ tilde - rho ] ) is to decide whether improving the model by adding planes will suffice , or shorter steps have to be forced by increasing @xmath153 .
the denominator in ( [ tilde - rho ] ) gives the model predicted progress @xmath159 at @xmath83 . on the other hand ,
the numerator @xmath160 gives the progress over @xmath35 we would achieve at @xmath83 , had we already known the cutting planes drawn at @xmath83 . due to aggregation we know that @xmath161 , so that @xmath162 , but values @xmath163 indicate that little to no progress is achieved by adding the cutting plane . in this case
we decide that the @xmath153-parameter must be increased to force smaller steps , because that reinforces the agreement between @xmath3 and @xmath155 . in the test ( [ dec ] )
we replace @xmath164 by @xmath165 for some fixed @xmath166 . if @xmath167 , then the quotient if far from 1 and we decide that adding planes has still the potential to improve the situation . in that event
we do not increase @xmath153 .
let us next consider the management of @xmath153 in the outer loop . since @xmath153 can only increase or stay fixed in the inner loop , we allow @xmath153 to decrease between serious steps @xmath168 , respectively , @xmath169 .
this is achieved by the test @xmath170 where @xmath171 is fixed . in other words ,
if at acceptance we have not only @xmath172 , but even @xmath173 , then we decrease @xmath153 at the beginning of the next inner loop @xmath174 , because we may trust the model . on the other hand ,
if @xmath175 at acceptance , then we memorize the last @xmath153-parameter used , that is @xmath176 at the end of the @xmath177th inner loop .
we should compare our management of the proximity control parameter @xmath153 with other strategies in the literature .
for instance mkel _
_ @xcite consider a very different management of @xmath153 , which is motivated by the convex case .
we are now ready to give our formal statement of algorithm [ algo1 ] . ' '' '' @xmath178 , @xmath179 , @xmath180 , @xmath181 . .
choose initial guess @xmath182 with @xmath183 and an initial matrix @xmath184 with @xmath185 .
fix memory control parameter @xmath186 such that @xmath187 .
put @xmath188 . .
at outer loop counter @xmath177 , stop if @xmath189 for some multiplier @xmath190 .
otherwise goto inner loop . .
put inner loop counter @xmath191 and initialize @xmath153-parameter using the memory element , i.e. , @xmath192 .
choose initial convex working model @xmath193 , possibly recycling some planes from previous sweep @xmath194 , and let @xmath195 .
at inner loop counter @xmath75 solve tangent program @xmath196 the solution is the new trial step @xmath197 .
check whether @xmath198 if this is the case put @xmath199 ( serious step ) , quit inner loop and goto step 8 .
if this is not the case ( null step ) continue inner loop with step 6 . .
build new convex working model @xmath200 based on null step @xmath197 by adding an exactness plane @xmath201 satisfying @xmath202 , a downshifted tangent @xmath203 , and the aggregate plane @xmath204 .
apply rule @xmath118 to avoid overflow .
build @xmath205 , and goto step 7 . .
compute @xmath206 put @xmath207then increase counter @xmath75 and continue inner loop with step 4 . .
update matrix @xmath208 , respecting @xmath209 and @xmath210 . then store new memory element @xmath211 increase @xmath212 if necessary to ensure @xmath213 .
if @xmath214 then re - set @xmath215 .
increase outer loop counter @xmath177 by 1 and loop back to step 2 . ' '' ''
in the convex cutting plane method @xcite unsuccessful trial steps @xmath83 are cut away by adding a tangent plane to @xmath3 at @xmath83 into the model . due to convexity
, the cutting plane is below @xmath3 and can therefore be used to construct an approximation ( [ working ] ) of @xmath3 . for nonconvex @xmath3 ,
cutting planes are more difficult to construct , but several ideas have been discussed .
we mention @xcite . in @xcite
we have proposed an axiomatic approach , which has the advantage that it covers the applications we are aware of , and allows a convenient convergence theory .
here we use this axiomatic approach in the convergence proof .
[ compare @xcite ] let @xmath3 be locally lipschitz .
a cutting plane oracle for @xmath3 on the set @xmath57 is an operator @xmath216 which , with every pair @xmath217 , @xmath35 a serious iterate in @xmath57 , @xmath218 a null step , associates an affine function @xmath219 , called the cutting plane at null step @xmath133 for serious iterate @xmath35 , so that the following axioms are satisfied : 1
. for @xmath220 we have @xmath221 and @xmath135 .
2 . let @xmath222 .
then there exist @xmath223 such that @xmath224 .
3 . let @xmath225 and @xmath226 .
then there exists @xmath227 such that + @xmath228 .
@xmath52 as we shall see , these axioms are aligned with the model axioms @xmath229 .
not unexpectedly , there is also a strict version of @xmath230 .
[ strict - oracle ] a cutting plane oracle @xmath216 for @xmath3 is called strict at @xmath5 if the following strict version of @xmath230 is satisfied : 1 .
suppose @xmath231 .
then there exist @xmath232 such that @xmath233 .
@xmath52 we now discuss two versions of the oracle which are of special interest for our applications .
[ model - based oracle ] suppose @xmath48 is a model of @xmath3
. then we can generate a cutting plane for serious iterate @xmath35 and trial step @xmath133 by taking @xmath234 and putting @xmath235 oracles generated by a model @xmath48 in this way will be denoted @xmath236 .
note that @xmath236 coincides with the standard oracle if @xmath3 is convex and @xmath67 , i.e. , if the convex @xmath3 is chosen as its own model . in more general cases ,
the simple idea of this oracle is that in the absence of convexity , where tangents to @xmath3 at @xmath133 are not useful , we simply take tangents of @xmath237 at @xmath133 . note that the model - based oracle @xmath236 is strict as soon as the model @xmath48 is strict .
@xmath52 [ standard oracle ] a special case of the model - based oracle is obtained by choosing the standard model @xmath56 . due to its significance for our present work
we call this the standard oracle . the standard cutting plane for serious step @xmath35 and null step @xmath133 is @xmath238 , where the clarke subgradient @xmath135 is one of those that satisfy @xmath239 .
the standard oracle is strict iff @xmath56 is strict .
as was observed before , this is for instance the case when @xmath3 is upper-@xmath0 .
note a specificity of the standard oracle : every standard cutting plane @xmath240 is also an exactness plane at @xmath35 .
@xmath52 [ downshifted tangents ] probably the oldest oracle used for nonconvex functions are downshifted tangents , which we define as follows . for serious iterate @xmath35 and null step @xmath133
let @xmath241 be a tangent of @xmath3 at @xmath133 .
that is , @xmath242
. then we shift @xmath243 down until it becomes useful for the model ( [ working ] ) .
fixing a parameter @xmath244 , this is organized as follows : we define the cutting plane as @xmath245 , where the downshift @xmath246 satisfies @xmath247_+ .\ ] ] in other words , @xmath248 , where @xmath249 .
note that this procedure aways satisfies axioms @xmath250 and @xmath251 , whereas axioms @xmath230 , respectively , @xmath252 , are satisfied if @xmath3 is lower-@xmath0 at @xmath5 .
in other words , see @xcite , for @xmath3 lower-@xmath0 this is an oracle , which is automatically strict .
@xmath52 motivated by the previous examples , we now define an oracle which works for both lower-@xmath0 and upper-@xmath0 .
[ modified downshift ] [ modified ] let @xmath35 be the current serious iterate , @xmath133 a null step in the inner loop belonging to @xmath35 .
then we form the downshifted tangent @xmath253 , that is , the cutting plane we would get from the downshift oracle , and we form the standard oracle plane @xmath254 , where the clarke subgradient @xmath69 satisfies @xmath255 . then we define @xmath256 in other words , among the two candidate cutting planes @xmath257 and @xmath240 , we take the one which has the larger value at the null step @xmath133 . note
that this is the oracle we use in our algorithm .
theorem [ theorem1 ] clarifies when this oracle is strict .
@xmath52 given an operator @xmath216 which with every pair @xmath217 of serious step @xmath35 and null step @xmath133 associates a cutting plane @xmath258 , we fix a constant @xmath259 and define what we call the upper envelope function of the oracle @xmath260 the crucial property of @xmath261 is the following suppose @xmath262 is a cutting plane oracle satisfying axioms @xmath263 .
then @xmath261 is a model of @xmath3 .
moreover , if the oracle satisfies @xmath252 , then @xmath261 is strict .
@xmath52 the proof can be found in @xcite .
we refer to @xmath261 as the upper envelope model associated with the oracle @xmath216 .
since in turn every model @xmath48 gives rise to a model - based oracle , @xmath236 , it follows that having a strict oracle and having a strict model are equivalent properties of @xmath3 .
note , however , that the model @xmath261 is in general not practically useful .
it is a theoretical tool in the convergence proof .
if we start with a model @xmath48 , then build @xmath236 , and go back to @xmath261 , we get back to @xmath48 , at least locally . on the other hand , going from an oracle @xmath216 to its envelope model @xmath261 , and then back to the model based oracle @xmath264 does _ not _ necessarily lead back to the oracle @xmath216 .
we are now in the position to check axiom @xmath130 .
all working models @xmath89 constructed in our algorithm satisfy @xmath265 .
in this section we state and prove the main result of this work and give several consequences .
[ theorem1 ] let @xmath3 be locally lipschitz and suppose for every @xmath17 , @xmath3 is either lower-@xmath0 or upper-@xmath0 at @xmath35 .
let @xmath182 be such that @xmath183 and @xmath266 is bounded .
then every accumulation point @xmath267 of the sequence @xmath154 of serious iterates generated by algorithm [ algo1 ] is a kkt - point of ( [ program ] ) .
the result will follow from ( * ? ? ? * theorem 1 ) as soon as we show that downshifted tangents as modified in example [ modified ] and used in the algorithm is a strict cutting plane oracle in the sense of definition [ strict - oracle ] .
the remainder of the proof is to verify this .
\1 ) let us denote cutting planes arising from the standard model @xmath56 by @xmath240 , cutting planes obtained by downshift as @xmath268 , and the true cutting plane of the oracle as @xmath269 .
then as we know @xmath270 if @xmath271 , and otherwise @xmath272 .
we have to check @xmath250 , @xmath252 , @xmath251 .
\2 ) the validity of @xmath250 is clear , as both oracles provide clarke tangent planes to @xmath3 at @xmath35 for @xmath220 .
\3 ) let us now check @xmath251 .
consider @xmath225 , and @xmath273 . here
@xmath274 is a null step at serious step @xmath275 . passing to a subsequence , we may distinguish case i , where @xmath276 for every @xmath177 , and case ii , where @xmath277 for every @xmath177 .
consider case i first .
let @xmath278 , where @xmath279 satisfies @xmath280 .
passing to yet another subsequence , we may assume @xmath281 , and upper semi - continuity of the clarke subdifferential gives @xmath135 .
therefore @xmath282 .
so here @xmath251 is satisfied with @xmath283 .
newt consider case ii .
here we have @xmath284 , where @xmath285 is a tangent to @xmath3 at @xmath274 with subgradient @xmath286 , and
@xmath287 is the corresponding downshift @xmath288_+.\ ] ] passing to a subsequence , we may assume @xmath281 , and by upper semi - continuity of @xmath289 we have @xmath242 . therefore @xmath290_+=:s$ ] , where uniform convergence @xmath291 occurs due to the boundedness of @xmath289 .
but now we see that @xmath292 is the downshift for the pair @xmath217 when @xmath242 is used .
hence @xmath293 , and since @xmath294 , we are done .
so again the @xmath295 in @xmath251 equals @xmath133 here .
\4 ) let us finally check axiom @xmath252 .
let @xmath296 be given .
we first consider the case when @xmath3 is upper-@xmath0 at @xmath35 .
we have to find @xmath232 such that @xmath297 as @xmath298 , and by the definition of the oracle , it clearly suffices to show @xmath299 . by spingarn @xcite , or daniilidis and georgiev @xcite , @xmath13 , which is lower-@xmath0 at @xmath35 , has the following property :
for every @xmath60 there exists @xmath300 such that for all @xmath301 and @xmath302 , @xmath303 taking the limit superior @xmath304 implies @xmath305 choosing @xmath306 , @xmath307 , @xmath308 , we can find @xmath223 such that @xmath309 , hence @xmath299 by the definition of @xmath310 . that settles the upper-@xmath0 case .
now consider the case where @xmath3 is lower-@xmath0 at @xmath35 .
we have to find @xmath232 such that @xmath297 as @xmath298 , and it suffices to show @xmath311 . since @xmath312 , where @xmath287 is the downshift @xmath313_+$ ] , and @xmath314 for some @xmath315 , it suffices to exhibit @xmath232 such that @xmath316 , or what is the same , @xmath317 . for
that it suffices to arrange @xmath318_+
\leq \epsilon_j \|y_j - x_j\|$ ] , because once this is verified , we get @xmath319_+ + c\|y_j - x_j\|^2 \leq ( \epsilon_j + c\|y_j - x_j\| ) \|y_j - x_j\| = : \widetilde{\epsilon}_j \|y_j - x_j\|$ ] . note again that by @xcite @xmath3 has the following property at @xmath35 : for every @xmath60 there exists @xmath45 such that @xmath320 for all @xmath302 .
dividing by @xmath321 and passing to the limit @xmath304 gives @xmath322 , using the fact that @xmath3 is locally lipschitz .
but for every @xmath242 , @xmath323 . using @xmath324 and taking @xmath307 , @xmath306 , this allows us to find @xmath232 such that @xmath325 .
substituting this above gives @xmath326 as desired .
that settles the lower-@xmath0 case .
in this section we discuss several technical aspects of the algorithm , which are important for its performance .
the stopping test in step 2 of the algorithm is stated in this form for the sake of the convergence proof . in practice we delegate stopping to the inner loop using the following two - stage procedure .
if the inner loop at serious iterate @xmath154 finds the new serious step @xmath327 such that @xmath328 then we decide that @xmath327 is optimal . in consequence , the @xmath329st inner loop will not be executed . on the other hand , if the inner loop has difficulties terminating and produces five consecutive null steps @xmath83 where @xmath330 or if a maximum number @xmath331 of allowed steps in the inner loop is reached , then we decide that @xmath154 is optimal . in our experiments we use , and @xmath332 . at the beginning of a new inner loop at serious step @xmath327
, we do not want to start building the working model @xmath333 from scratch .
it is more efficient to recycle some of the planes @xmath334 in the latest working model @xmath335 . in the convex cutting plane method , this is self - understood , as cutting planes are affine minorants of @xmath3 , and can at leisure stay on in the sets @xmath336 at all times @xmath337 . without convexity , we need the following recycling procedure : given a plane @xmath338 in the latest set @xmath339 , we form the new downshifted plane @xmath340 where the downshift is organized as @xmath341_+.\ ] ] in other words , we treat @xmath342 like a tangent to @xmath3 at null step @xmath154 with respect to the serious step @xmath327 in the downshift oracle .
we put @xmath343 and we accomodate @xmath344 at the beginning of the @xmath329st inner loop .
the interface behavior of laminated composite materials is modeled by a non - monotone multi - valued function @xmath345 , characteristic of the interlayer adhesive placed at the contact boundary @xmath346 . in more precise terms
, @xmath345 is the physical law which holds between the normal component @xmath347 of the stress vector and the relative displacement @xmath348 , or _ jump _
, between the upper and lower boundaries .
a typical law @xmath345 for an interlayer adhesive is shown in figure [ dellaw ] ( left ) . in the material sciences ,
the knowledge of @xmath345 is crucial for the understanding of the basic failure modes of the composite material .
the adhesive law @xmath345 is usually determined experimentally using the double cantilever beam test @xcite or other destructive testing methods .
the result of a typical experiment is shown schematically in figure [ contaminated ] from @xcite , where three probes with different levels of contamination have been exposed .
while the intact material shows stable propagation of the crack front ( dashed curve ) , the 10% contaminated specimen shows a typical zig - zag profile ( bold solid curve ) , indicating unstable crack front propagation .
indeed , when reaching the critical load @xmath349n , the crack starts to propagate .
since by the growth of the crack - elongation , the compliance of the structure increases , the crack propagation slows down and the crack is `` caught '' , i.e. , stops at @xmath350 mm and the load @xmath351 in the structure drops from @xmath349n to @xmath352n .
thereafter , due to the continuously increased load , the crack starts again to propagate until reaching another critical load level at @xmath353 and @xmath354 mm .
this phenomenon occurs five to six times , as seen in figure [ contaminated ] . , leading to an upper-@xmath0 objective .
right image shows non - monotone friction law , leading to a lower-@xmath0 objective .
[ dellaw ] ] the 50% contaminated specimens ( dotted curve ) shows micro - cracks that appear at a finer level and are not visible in the figure [ contaminated ]
. the lower level of the adhesive energy , which is represented by the area below the load - displacement curve , indicates now that this specimen is of minor resistance .
even though the displacement @xmath355 in figure [ contaminated ] can only be measured at the crack tip , in order to proceed one now
_ stipulates _ the law @xmath345 all along @xmath356 by assuming that the normal stresses @xmath357 follow the measured behavior @xmath358 under this hypothesis one now solves the variational inequality for the unknown displacement field @xmath359 , and then validates ( [ law1 ] ) .
note that @xmath357 is the _ truly _ relevant information , as it indicates the action of the destructive forces along @xmath346 , explaining eventual failure of the composite . in current practice in the material sciences ,
this information can not be assessed by direct measurement , and is therefore estimated by heuristic formulae @xcite .
our approach could be interpreted as one such estimation technique based on mathematical modeling . within the framework of plane linear elasticity we consider a symmetric laminated structure with an interlayer adhesive under loading ( see fig . [ model ] ) .
because of the symmetry of the structure , it suffices to consider only the upper half of the specimen , represented by @xmath360 .
the lipschitz boundary @xmath361 of @xmath362 consists of four disjoint parts @xmath363 , @xmath346 , @xmath364 and @xmath365 .
the body is fixed on @xmath363 , i.e. , @xmath366 on @xmath367 the traction forces @xmath368 are constant and given as @xmath369 the part @xmath370 is load - free .
we adopt standard notation from linear elasticity and introduce the bilinear form of linear elasticity @xmath371 where @xmath372 is the displacement vector , @xmath373 the linearized strain tensor , and @xmath374 the stress tensor . here , @xmath375 is the elasticity tensor with symmetric positive @xmath376 coefficients .
the bilinear form is symmetric and due to the first korn inequality , coercive .
the linear form @xmath377 is defined by @xmath378 on the contact boundary @xmath346 we have the unilateral constraint @xmath379 and we apply the non - monotone multi - valued adhesive law @xmath380 here @xmath381 , where @xmath382 is the outward unit normal vector to @xmath346 . a typical non - monotone law @xmath383 for delamination , describing the behavior of the adhesive , is shown in fig .
[ dellaw ] .
this law is derived from a nonconvex and a nonsmooth locally lipschitz super - potential @xmath177 expressed in terms of a minimum function . in particular
, @xmath384 is a minimum of four convex quadratic and one linear function .
we also assume that tangental traction can be neglected on @xmath346 , i.e. , @xmath385 . the weak formulation of the delamination problem is then given by the following hemivariational inequality : find @xmath386 such that @xmath387 where @xmath388 is the clarke directional derivative of @xmath384 at @xmath389 in direction @xmath55 , @xmath6 is the nonempty , closed convex set of all admissible displacements defined by @xmath390 contained in the function space @xmath391 the potential energy of the problem is @xmath392 where @xmath393 defined by @xmath394 is the term responsible for the nonsmoothness . using the potential energy , the hemivariational inequality ( [ del ] ) can be transformed to the following nonsmooth , nonconvex constrained optimization problem of the form ( [ program ] ) @xmath395 where the objective is upper-@xmath0 , because the super - potential @xmath384 is a minimum .
in particular , we have an objective of the form ( [ structure ] ) , where the smooth part @xmath15 comprises @xmath396 , while the nonsmooth part @xmath397 has the form ( [ structure ] ) with a finite index set @xmath26 once the boundary integral is suitably parametrized .
according to the existence theory in @xcite , problem ( [ del2 ] ) has at least one clarke critical point @xmath398 satisfying the necessary optimality condition @xmath399 where @xmath400 is the normal cone to @xmath6 at @xmath401 , and vice versa , by a result in @xcite every critical point of @xmath402 on @xmath6 is a solution of ( [ del ] )
( see also @xcite ) .
n is reached , the crack starts to propagate .
but then the propagation speed slows down , since by the crack the compliance of the specimen increases , and the crack is `` caught '' at @xmath350 mm .
the load @xmath351 drops from @xmath349n to @xmath352n .
then , by the constantly applied traction force , there is a linear growth of the load @xmath351 from @xmath403n to the critical load @xmath404n , where the crack propagates again and stops at @xmath354 mm , with the load now reduced to @xmath405n .
the 50% contaminated specimen exhibits micro - cracks not visible at the chosen scale . [
contaminated ] ] we consider a regular triangulation @xmath406 of @xmath57 , where we first divide @xmath57 into small squares of size @xmath407 and then each square by its diagonal into two triangles . to approximate @xmath408 and @xmath6 we use a piecewise linear finite element approximation and set @xmath409 @xmath410 similar to low order finite element approximations of nonsmooth convex contact problems @xcite
, we use the trapezoidal quadrature rule to approximate the functional @xmath411 by @xmath412 , \ ] ] where we are summing over the nodes @xmath413 on the contact boundary @xmath414 , with @xmath415 being the neighbor of node @xmath413 on @xmath346 in the sense of integration .
this can be regrouped as @xmath416 with appropriate weights @xmath417 .
here , @xmath418 is the set of zig - zags in the graph of @xmath345 .
the bundle algorithm is applied to minimize the discrete functional @xmath419 introducing an index set @xmath420 for the nodes @xmath413 on the contact boundary @xmath421 , we may pull out the minimum from under the sum , which leads to the expression @xmath422 this is the discrete version of ( [ struct ] ) , where @xmath423 is the smooth term @xmath15 , and @xmath424 the nonsmooth part . while computation of clarke subgradients is straightforward here , we still have to explain how the matrix @xmath425 in the second - order working model ( [ second ] ) is chosen . discretizing the quadratic form of linear elasticity as @xmath426 with the symmetric stiffness matrix @xmath427 , and observing
that @xmath428 is linear , we choose @xmath429 , where @xmath430 is one of those indices , where the minimum @xmath431 is attained .
for convergence of the lowest - order finite element approximation used here we refer to the results in @xcite .
higher - order approximations with no limitation in the polynomial degree , which lead to nonconforming approximation of unilateral constraints , have only recently been analyzed for monotone contact problems , see @xcite .
we present numerical results obtained in a delamination simulation with modulus of elasticity @xmath432 gpa and poisson ratio @xmath433 corresponding to a steel specimen . in all examples we use the benchmark model of @xcite with geometrical characteristics @xmath434 in [ mm ] and thickness @xmath435 mm .
we apply our bundle method to ( [ del2 ] ) and compare the results to those obtained by the regularization technique in @xcite .
all computations use piecewise linear functions and the discretization @xmath436 corresponding to @xmath437 cm . in this case , the number of the unknowns in the discrete problem ( [ discretej ] ) is @xmath438 .
for 5 different values of @xmath439 .
right image shows vertical component of reactive force along contact boundary for same 5 scenarios.,title="fig:",width=294,height=173 ] for 5 different values of @xmath439 .
right image shows vertical component of reactive force along contact boundary for same 5 scenarios.,title="fig:",width=294,height=173 ] for 5 different values of @xmath439 .
right image shows vertical component of reactive force along contact boundary for same 5 scenarios.,title="fig:",width=294,height=177 ] for 5 different values of @xmath439 .
right image shows vertical component of reactive force along contact boundary for same 5 scenarios.,title="fig:",width=294,height=173 ] .regularization .
vertical displacement [ mm ] at 4 intermediate points for same 5 scenarios .
[ cols="^,^,^,^,^",options="header " , ]
we have presented a bundle method based on the mechanism of downshifted tangents which is suited to optimize upper- and lower-@xmath0 functions .
our method allows to integrate second - order information , if available , and gives a convergence certificate in the sense of subsequences . every accumulation point of
the sequence of serious iterates with an arbitrary starting point is critical .
we have successfully applied our method to a delamination problem arising in the material sciences , where upper-@xmath0 functions have to be minimized .
results obtained by optimization were compared to results obtained by the regularization technique of @xcite , and both methods are in good agreement .
the authors thank h .- j .
gudladt for many useful discussions .
the authors were partially supported by bayerisch - franzsisches hochschulzentrum ( bfhz ) .
99 m. wetzel , j. holtmannsptter , h .- j .
gudladt , j. v. czarnecki : sensitivity of double cantilever beam test to surface contamination and surface pretreatment . international journal of adhesion & adhesives , vol .
46 , 114 - 121 ( 2013 ) r. mifflin : a modification and an extension of lemarchal s algorithm for nonsmooth minimization .
study 17 , 77 - 90 ( 1982 ) c. lemarchal : bundle methods in nonsmooth optimization . in nonsmooth optimization ( proc .
iiasa workshop , laxenburg , 1977 ) , pp .
79 - 102 , iiasa proc .
ser . , 3 , pergamon , oxford - elmsford , n.y . , 1978 .
c. lemarchal , c. sagastizbal : variable metric bundle methods : from conceptual to implementable forms .
programming 76 ( 1997 ) , no .
b , 393 - 410 . : the bt - algorithm for minimizing a nonsmooth functional subject to linear constraints , in nonsmooth optimization and related topics , f. h. clarke , v. f. demyanov , f. gianessi ( eds . ) , plenum press ( 1989 ) : a version of the bundle idea for minimizing a nonsmooth function : conceptual idea , convergence analysis , numerical results , siam j. optim . 2 , 121 - 152 ( 1992 ) d. noll : cutting plane oracles to minimize nonsmooth nonconvex functions .
set - valued var .
18 ( 3 - 4 ) , 531 - 568 ( 2010 ) d. noll , o. prot , a. rondepierre : a proximity control algorithm to minimize nonsmooth nonconvex functions .
pacific j. optim . 4 ( 3 ) , 569 - 602 ( 2008 ) d. alazard , m. gabarrou , d. noll : design of a flight control architecture using a nonconvex bundle method .
control sign .
25 ( 2 ) , 257 - 290 ( 2013 ) d. noll : convergence of nonsmooth descent methods using the kurdyka - ojasiewicz inequality .
theory appl .
( doi ) 10.1007/s10957 - 013 - 0391 - 8 .
mkel , m. miettinen , l. lukan , j. vlek : comparing nonsmooth nonconvex bundle methods in solving hemivariational inequalities , journal of global optimization 14 ( 2 ) , 117 - 135 ( 1999 ) .
m. miettinen , m.m .
mkel , j. haslinger : on numerical solution of hemivariational inequalities by nonsmooth optimization methods , journal of global optimization 6 ( 4 ) , 401 - 425 ( 1995 ) .
l. lukan , j. vlek : a bundle - newton method for nonsmooth unconstrained minimization , math .
83 , 373 - 391 ( 1998 ) j. haslinger , m. miettinen , p.d .
panagiotopoulos : finite element methods for hemivariational inequalities , kluwer academic publishers ( 1999 ) j. czepiel : proximal bundle method for a simplified unilateral adhesion contact problem of elasticity , schedae informaticae 20 , 115 - 136 ( 2011 ) l. nesemann , e.p .
stephan : numerical solution of an adhesion problem with fem and bem , appl .
62 ( 5 ) , 606 - 619 ( 2012 ) m. kovara , a. mielke , t. roubek : a rate - independent approach to the delamination problem , math .
solids 11 , no . 4 , 423 - 447 ( 2006 ) t. roubek , v. mantic , panagiotopoulos , c.g . : a quasistatic mixed - mode delamination model , discrete contin .
s 6 , no . 2 , 591 - 610 ( 2013 ) j. e. spingarn : submonotone subdifferentials of lipschitz functions .
264 , 77 - 89 ( 1981 ) a. daniilidis , p. georgiev : approximate convexity and submonotonicity . j. math .
291 , 117 - 144 ( 2004 ) r. t. rockafellar , r. j - b .
wets : variational analysis .
springer verlag ( 2004 ) a. daniilidis , j. malick : filling the gap between lower-@xmath0 and lower-@xmath71 functions .
journal of convex analysis 12(2 ) , 2005 , pp .
315 329 .
r. a. poliquin , r. t. rockafellar : prox - regular functions in variational analysis , trans .
348 ( 5 ) , 1805 - 1838 ( 1996 ) k.c .
kiwiel : an aggregate subgradient method for nonsmooth convex minimization , math .
programming 27 , 320 - 341 ( 1983 ) j. cullum , w.e .
donath , p. wolfe : the minimization of certain nondifferential sums of eigenvalues of symmetric matrices .
stud . 3 , 35 - 55 ( 1975 )
p. apkarian , d. noll , o. prot : a trust region spectral bundle method for nonconvex eigenvalue optimization , siam j. optim .
10 ( 1 ) , 281 - 306 ( 2008 ) a. ruszczyski : nonlinear optimization , princeton university press ( 2006 ) j .- b .
hiriart - urruty , c. lemarchal : convex analysis and minimization algorithms , vol .
i and ii : advanced theory and bundle methods , vol .
306 of grundlehren der mathematischen wissenschaften , springer verlag , new york , heidelberg , berlin ( 1993 ) w. l. hare , c. sagastizabal : computing proximal points of nonconvex functions , math .
programming series b 116 , 221 - 258 ( 2009 ) z. naniewicz , p.d .
panagiotopoulos : mathematical theory of hemivariational inequalities and applications , new york ( 1995 ) .
n. ovcharova : regularization methods and finite element approximation of hemivariational inequalities with applications to nonmonotone contact problems , phd thesis , universitt der bundeswehr mnchen , cuvillier verlag , gttingen ( 2012 ) .
n. ovcharova , j. gwinner : a study of regularization techniques of nondifferentiable optimization in view of application to hemivariational inequalities , accepted for publication in jota , jota - d-13 - 00163 ovcharova , n. , gwinner .
j : on the regularization method in nondifferentiable optimization applied to hemivariational inequalities , constructive nonsmooth analysis and related topics , springer , 59 - 70 ( 2013 ) .
baniotopoulos , j. haslinger , z. morvkov : contact problems with nonmonotone friction : discretization and numerical realization , comput .
40 , 157 - 165 ( 2007 ) c.c .
baniotopoulos , j. haslinger , z. morvkov : mathematical modeling of delamination and nonmonotone friction problems by hemivariational inequalities , applications of mathematics 50 ( 1 ) , 1 - 25 ( 2005 ) s. carl , v.k .
le , d. motreanu : nonsmooth variational problems and their inequalities , springer ( 2007 ) r. glowinski : numerical methods for nonlinear variational problems , springer , new york ( 1984 ) d. goeleven , d. motreanu , y. dumont , m. rochdi , m. : variational and hemivariational inequalities : theory , methods and applications , vol .
i : unilateral analysis and unilateral mechanics , vol .
ii : unilateral problems , kluwer ( 2003 ) j. gwinner : finite - element convergence for contact problems in plane linear elastostatics , quarterly of applied mathematics , vol . 50 , 11 - 25 ( 1992 ) j. gwinner : hp - fem convergence for unilateral contact problems with tresca friction in plane linear elastostatics , j. comput . appl .
254 , 175 - 184 ( 2013 ) p.d .
panagiotopoulos : hemivariational inequalities .
applications in mechanics and engineering , berlin , springer ( 1993 ) p.d .
panagiotopoulos : inequality problems in mechanics and application .
convex and nonconvex energy functions , basel , birkhuser ( 1998 ) m. sofonea , a. matei : variational inequalities with applications , springer ( 2009 )
|
delamination is a typical failure mode of composite materials caused by weak bonding .
it arises when a crack initiates and propagates under a destructive loading .
given the physical law characterizing the properties of the interlayer adhesive between the bonded bodies , we consider the problem of computing the propagation of the crack front and the stress field along the contact boundary .
this leads to a hemivariational inequality , which after discretization by finite elements we solve by a nonconvex bundle method , where upper-@xmath0 criteria have to be minimized . as this is in contrast with other classes of mechanical problems with non - monotone friction laws and in other applied fields , where criteria are typically lower-@xmath0 ,
we propose a bundle method suited for both types of nonsmoothness .
we prove its global convergence in the sense of subsequences and test it on a typical delamination problem of material sciences .
* key words . *
composite material @xmath1 delamination @xmath1 crack front propagation @xmath1 hemivariational inequality @xmath1 clarke directional derivative @xmath1 nonconvex bundle method @xmath1 lower- and upper-@xmath0 function @xmath1 convergence .
institut de mathmatiques , universit de toulouse , france ] institute of mathematics , department of aerospace engineering , universitt der bundeswehr mnchen , germany ]
|
the magnetic properties of transition metal compounds with @xmath2 valence electrons have been one of the central research themes in solid - state physics for the past three decades . in view of the interplay between magnetism and high - temperature superconductivity ,
particular attention has been focused on oxides and arsenides .
recently , the electronic structure and ordering phenomena of transition metal compounds with @xmath4 and @xmath5 valence electrons ( such as ruthenates and iridates ) have also captured much attention .
the electronic correlations in these materials are generally weaker than those of their @xmath2 counterparts , while the spin - orbit coupling is stronger .
the quantitative description of the influence of these parameters on the electronic phase behavior of @xmath6-electron compounds is an important topic of current research .
here we report a detailed investigation of the magnetic properties of pdas@xmath0o@xmath1 , a recently synthesized orosel electrically insulating compound with a magnetic lattice of pd@xmath7 ions in the electron configuration @xmath8 .
we compare our results to the isostructural compound nias@xmath0o@xmath1 , which is based on ni@xmath9 ions with the same number of electrons in the @xmath2-shell .
pdas@xmath0o@xmath1 crystallizes in the pbsb@xmath0o@xmath1 structure with pd@xmath9 and as@xmath10 ions segregated into different layers ( fig .
the octahedral coordination of pd@xmath9 in this structure is unusual , because divalent palladium shows a strong preference for square - planar coordination , which is associated with a diamagnetic ground state .
only a few examples of sixfold - coordinated pd@xmath9 compounds are known , including the ambient- and high - pressure polymorphs of pdf@xmath0 as well fluoro - palladates of composition _ _ m__pdf@xmath11 ( _ m _
= ca , cd , hg ) and cspd@xmath0f@xmath12 .
@xcite these compounds are paramagnetic at high temperatures and tend to order antiferromagnetically upon cooling . in accord with this trend ,
magnetic susceptibility measurements on pdas@xmath0o@xmath1 showed paramagnetic behavior at room temperature , and an antiferromagnetic phase transition at the nel temperature @xmath13 k. @xcite this behavior is qualitatively analogous to the one of the isostructural 3__d__-homologues mnas@xmath0o@xmath1 , coas@xmath0o@xmath1 , and nias@xmath0o@xmath1 , which also show antiferromagnetic ordering with @xmath14 , 20 , and 30 k , respectively .
@xcite however , the much higher nel temperature of pdas@xmath0o@xmath1 is surprising , especially because the pdo@xmath1-octahedra do not share vertices , edges or faces .
the exchange paths connecting neighboring pd@xmath9 ions are therefore long and involve at least two bridging oxygen sites . in order to elucidate the microscopic origin of this surprising behavior , we have used neutron diffraction to determine the magnetic structure of pdas@xmath0o@xmath1 , which turned out to be identical to that of the 3__d__-homologues ( section ii ) .
this implies similar networks of exchange bonds in both sets of compounds .
we employed density - functional calculations to obtain insights into the electronic structure of pdas@xmath0o@xmath1 and nias@xmath0o@xmath1 , and specifically into the origin of the exchange paths ( which turned out to be hopping via as dimers ) and of the large amplitude difference of the primary exchange interaction parameters ( section iii ) . a model based on these interactions
yields an excellent description of the magnetic susceptibilities of both compounds .
as@xmath15o@xmath16 ( @xmath17 = pd , ni ) showing the three - dimensional network of @xmath18 and as@xmath19 octahedra .
the brown ( dark grey ) colored balls denote @xmath17 atoms , and the cyan ( light grey ) colored balls represent as atoms .
small black colored balls at the corners of the octahedra are o atoms . ] o@xmath1 . ]
a powder sample of pdas@xmath0o@xmath1 of weight @xmath20 g was prepared using the starting materials pdo ( 99.9 % metals basis , alfa aesar ) and as@xmath0o@xmath12 ( 99.9 % , alfa aesar ) in the molecular ratio 1 : 1.1 as described earlier .
@xcite the mixed powder was pressed into pellets and dried in evacuated silica tubes for 12 h at 373 k. then the evacuated silica tubes were heated up to 973 k with a rate of 100 k / h .
the hygroscopic and air sensitive powder of pdas@xmath0o@xmath1 was obtained after an annealing process of about 100 hours .
measurements of the magnetic susceptibility and specific heat were carried out in the temperature range between 5 and 300 k ( fig .
both quantities show anomalies indicative of antiferromagnetic ordering of the pd sublattice at 140 k , in agreement with prior work .
orosel in order to investigate the crystal and magnetic structure of pdas@xmath0o@xmath1 , a neutron diffraction experiment was carried out at the research reactor frm - ii in garching .
neutron powder patterns were collected with the instrument spodi at 5 k and 200 k in the @xmath21 range 4@xmath22160@xmath22 .
this instrument uses a germanium monochromator ( reflection 551 ) selecting the neutron wavelength @xmath23 .
the refinements of the crystal and magnetic structure were carried out with the program fullprof .
@xcite we used the nuclear scattering lengths _
b_(pd ) = 5.91 fm , _
b_(as ) = 6.58 fm and _ b_(o ) = 5.805 fm .
sears the magnetic form factors of the magnetic ions were taken from ref . .
the crystal structure of pdas@xmath0o@xmath1 was recently refined from x - ray powder diffraction data in the trigonal pbsb@xmath0o@xmath1-type structure ( space group _ _ p__@xmath241__m _ _ , no .
162 ) , where the pd , as and o atoms are in the wyckoff positions 1__a__(0,0,0 ) , 2__d__(@xmath25,@xmath26,@xmath27 ) and 6__k__(_x_,0,_z _ ) , respectively .
@xcite the same space group was found earlier for the compounds mnas@xmath0o@xmath1 , coas@xmath0o@xmath1 and nias@xmath0o@xmath1 containing 3__d__-metal ions .
@xcite from our neutron powder diffraction data taken at the lower temperatures 5 and 200 k ( fig .
3 ) the trigonal space group _ _ p__@xmath241__m _ _ was confirmed . for the rietveld refinements we used data in the extended @xmath21 range from 4@xmath22 up to 146@xmath22 .
a total of 14 parameters was refined : an overall scale factor , five profile function parameters , the zero point , two lattice constants , the positional parameters _ x _ and _ z _ of the oxygen atom as well as three isotropic thermal parameters .
the powder sample contained small amounts of the binary oxide pdo , which crystallizes in the tetragonal space group @xmath28 .
@xcite therefore the overall scale factor of pdo was additionally allowed to vary during the refinements . in table
i the results of the refinements are compared with those of the x - ray study carried out earlier at room temperature .
@xcite here it can be seen that the positional parameters of the oxygen atoms determined at 5 and 200 k are in good agreement , indicating that the structural changes between the magnetically ordered and the paramagnetic states are weak .
only a slight reduction of 0.0036 ( about 6 @xmath29 ) could be observed for the pd - o - bond length in the pdo@xmath1-octahedra .
in contrast , the distances between the as and o - atoms are practically unchanged ( table i ) .
the value _ d_(as - o ) = 1.8281(6 ) ( at 200 k ) found for pdas@xmath0o@xmath1 is in very good agreement with the values of other arsenates containing 3__d__-metal ions : _
d_(as - o ) = 1.827(4 ) ( nias@xmath0o@xmath1 ) , _ d_(as - o ) = 1.830(3 ) , ( coas@xmath0o@xmath1 ) , _ d_(as - o ) = 1.826(2 ) ( mnas@xmath0o@xmath1 ) , and _
d_(as - o ) = 1.826(1 ) ( cdas@xmath0o@xmath1 ) .
@xcite all of these values are in agreement with _ d_(as - o ) = 1.82 calculated for an aso@xmath1-octahedron given by shannon .
@xcite in table i it can be seen that the structural parameters obtained at 200 and 300 k show relatively large discrepancies , despite the fact that both data sets were collected in the paramagnetic phase .
this can be ascribed to the larger scattering power of the o - atoms in neutron diffraction , with the consequence that the o - positions can be determined more reliably .
furthermore , the shortest oxygen contact _ d_(o - o ) = 2.308(3 ) was found to be implausibly short in the x - ray study .
@xcite from our neutron diffraction study we found the larger values _
d_(o - o ) = 2.3757(12 ) at 5 k and _ d_(o - o ) = 2.3726(12 ) at 200 k , respectively .
the value _ d_(o - o ) = 2.410(3 ) found for nias@xmath0o@xmath1 is slightly larger , @xcite while the as - o - bond lengths _ d_(as - o ) = 1.827(4 ) ( nias@xmath0o@xmath1 ) and _ d_(as - o ) = 1.8281(6 ) ( pdas@xmath30o@xmath1 ) are practically the same in both compounds .
the cell volume of the ni compound ( @xmath31 @xmath32 ) is much smaller than the one of the pd compound ( @xmath33 @xmath32 ) .
this is due to fact that the ionic radius of pd@xmath9 is larger than that of ni@xmath9 . in order to keep the as - o - bond lengths almost constant in the aso@xmath1-octahedra ,
the bond angle @xmath34(o - as - o ) increases from 169.91(3)@xmath22 in pdas@xmath0o@xmath1 ) to 173.23(12)@xmath22 in nias@xmath0o@xmath1 .
.results of rietveld refinements of the neutron powder diffraction data ( @xmath35 ) for the nuclear structure of pdas@xmath30o@xmath1 at 5 and 200 k. the lattice constants , positional and isotropic thermal parameters as well as the bond distances and angles within the aso@xmath1- and pdo@xmath1-units are compared with the values obtained earlier at room temperature from x - ray powder diffraction data ( @xmath36 ) .
@xcite the residual _
_ r__@xmath37 of the refinement of the crystal structure is defined as @xmath38 ) was calculated with intensities rather than structure factors . @xcite [ cols="<,<,<,<",options="header " , ] the magnetic exchange interaction , @xmath39 , can be expressed in general as a sum of antiferromagnetic and ferromagnetic contributions . in the limit of large coulomb correlation ,
typically valid for late transition metal elements like pd or ni , the antiferromagnetic contribution is related to the hopping integral @xmath40 by the second - order perturbation relation @xmath39 @xmath41 , where @xmath42 is the effective on - site coulomb repulsion defined for the wannier orbitals .
considering the hopping integrals in table iii , the contribution from the @xmath43 term in the magnetic exchange gives rise to a factor two difference between the pd and ni compounds .
the estimate of @xmath42 , unlike that of the hopping integral @xmath40 , is a rather delicate issue . in principle
, one should compute @xmath42 for the wannier orbitals shown in figure 7 . however lacking a prescription to do so , we computed the @xmath42 values corresponding to ni-@xmath6 and pd-@xmath6 partial waves , which were truncated outside the atomic spheres defined around an ni or pd sites .
the calculations were carried out within the framework of the constrained dft scheme @xcite .
the @xmath42 values calculated in this way were @xmath44 ev and @xmath45 ev .
but since the wannier @xmath46 orbitals are far more delocalized than the truncated partial waves , and more so for pd than for ni , these @xmath42 values should be significantly reduced , and more so for pd than for ni .
this could conceivably lead to the factor @xmath47 needed to bring our theoretical @xmath48 estimate for the nel temperatures of the two compounds into agreement with the measured ratio of 4.7 .
taking into account only the dominant hopping integrals provided by the @xmath49mto - downfolding calculation , a spin model can be defined in terms of the six pairs of 3rd - nearest - neighbor magnetic interactions , all of size @xmath50 , obtained by summing over the squares of the @xmath46hopping integrals between the 3rd - nearest neighbors .
based on this model , we have calculated the magnetic susceptibility by considering the following spin-1 hamiltonian : @xmath51 this model was solved by the quantum monte carlo method @xcite on a @xmath52 lattice .
the primary interaction @xmath50 , and the effective land @xmath53-factors were obtained by fitting to the experimental susceptibility .
as shown in fig .
8 , the calculated and measured susceptibilities agree very well .
the optimal values of the @xmath53-factor and the exchange parameter @xmath50 , were found to be respectively 2.38 and 62 k for the pd compound , and 2.48 and 13 k for the ni compound .
we find that the @xmath53-factors are larger than the spin - only value of 2 , in accord with the discussion in section ii.c above .
the fit of the susceptibilities yields a value of 5 for the ratio of the predominant magnetic exchange parameters of pdas@xmath0o@xmath1 and nias@xmath0o@xmath1 , in good agreement with the ratio of the magnetic transition temperatures .
o@xmath1 ( left panel ) and pdas@xmath0o@xmath1 ( right panel ) .
the circles correspond to experimental data ( taken from ref . for nias@xmath0o@xmath1 ) , and the solid lines correspond to calculated data based on the model described in the text . ]
using a combination of susceptibility and neutron diffraction measurements , we have developed a comprehensive experimental description of the magnetic properties of the newly synthesized antiferromagnet pdas@xmath0o@xmath1 .
density functional theory has provided a detailed understanding of the magnetic bond network of this compound , as well as a semi - quantitative explanation of the large enhancement of the magnitude of its primary exchange interaction parameters compared to its @xmath2 homologue nias@xmath0o@xmath1 .
this approach may prove useful for research on other pd compounds including the recently discovered @xcite ferromagnet pds@xmath0o@xmath54 , and for comparative studies of materials with @xmath4 valence electrons and their @xmath2-electron counterparts in general .
we are grateful to the referee for constructive criticism , to m. hlzel and a. senyshyn for help with the neutron diffraction measurements at the frm - ii , and to o. jepsen for performing the @xmath42 calculations .
tsd and oka acknowledge the mpg - india partner group program and the indo - eu research network rp7 monami for support .
|
the crystal structure and magnetic ordering pattern of pdas@xmath0o@xmath1 were investigated by neutron powder diffraction .
while the magnetic structure of pdas@xmath0o@xmath1 is identical to the one of its isostructural @xmath2-homologue nias@xmath0o@xmath1 , its nel temperature ( 140 k ) is much higher than the one of nias@xmath0o@xmath1 ( 30 k ) .
this is surprising in view of the long distance and indirect exchange path between the magnetic pd@xmath3 ions .
density functional calculations yield insight into the electronic structure and the geometry of the exchange - bond network of both pdas@xmath0o@xmath1 and nias@xmath0o@xmath1 , and provide a semi - quantitative explanation of the large amplitude difference between their primary exchange interaction parameters .
|
non equilibrium transport between superconductors with a dc voltage bias gives rise to a subgap structure in the current voltage characteristics which can be described in terms of multiple andreev reflections ( mar)@xcite . indeed , it has been understood since the sixties@xcite that when the bias potential between two superconductors is smaller than the superconducting energy gap , electrons have to be transferred in bunches in order to satisfy energy requirements .
the calculation of the current in the presence of such mar processes can proceed along several directions .
early work@xcite considered a formulation of transport in terms of transmission probabilities rather than amplitudes . during the last decade or so , mar processes have been studied theoretically in the coherent regime for point contacts , using either scattering theory@xcite or microscopic tight binding hamiltonians@xcite .
the coherent current which flows between the two superconductors is then time dependent : it contains all harmonics of the josephson frequency . of particular interest in ref .
was the fact that in addition to the dc current , the cosine and sine harmonics of the current were computed .
these harmonics also exhibit structures at the mar onsets , and they allow for some additional diagnosis at low voltages : the amplitude of the sine first harmonic ( sfh ) at zero voltage corresponds to the critical current in the josephson ( zero bias ) limit , while the cosine first harmonic ( cfh ) vanishes . on the experimental side , in the context of mesoscopic physics
, pioneering experiments were performed on atomic point contacts for the current@xcite as well as for the noise@xcite . at the same time
, samples containing a diffusive normal metal sandwiched between superconducting leads were studied @xcite- a regime which also triggered theoretical activity .
@xcite given the recent interest in nanophysics for studying systems with reduced size ( such as quantum dots and molecules ) , mar through a quantum dot described as a resonant level without interactions has been addressed @xcite .
it was found that the position of the resonance bears strong consequences on the dc current voltage characteristics .
so far however , little is known on the harmonics of the supercurrent .
experiments on non equilibrium supercurrent through quantum dots have recently been performed , pointing out the effect of size quantization and in some cases of resonances , attributed to coulomb interactions@xcite .
so far mar transport calculations have been focused either on the incoherent or coherent cases .
the crossover regime was discussed for a chaotic dot junction where an external magnetic field serves as a dephasing factor@xcite . in mesoscopic physics ,
the current ( and noise ) which flows between a source and a drain can be modified if one inserts a voltage probe between the two@xcite .
the probe voltage can be adjusted so that no net electrons flow through it , electrons which enter this probe loose their phase coherence because they enter in contact with an electron reservoir . in mesoscopic transport
, such probes can be used as a way to mimic the effect of inelastic scattering / decoherence on transport between the source and the drain .
one can thus model the crossover between the two regimes by adjusting the degree of decoherence by varying the coupling to such voltage probes . at the same time , experimentally one could also construct actual nanostructures which contain a controlled connection to such voltage - like probes . in this paper , we address key issues associated with mar for a device consisting of two superconductor connected to a quantum dot / resonant level , in the absence and in the presence of a normal metal lead connected to the dot . to achieve this
, we will choose a hamiltonian formulation and we will use a keldysh green s function approach where the leads are effectively integrated out .
first , we will study the supercurrent harmonics of an isolated quantum dot .
we will first show under which conditions the regimes of a point contact and of a resonant level are recovered , and then discuss the properties of the supercurrent harmonics for the case of the resonant level ; we will show that these properties differ strongly from those of a quantum point contact .
next , we will introduce a normal probe which is directly connected to the dot , and we will monitor the full crossover from the coherent regime to the fully incoherent regime .
we will also compute the ( equilibrium ) josephson current independently , and we will check whether its first harmonic corresponds to the extrapolated amplitude of the sfh at zero bias . note that the setup is similar to that of ref .
, however , the attention there was put at the non - equilibrium josephson effect under current injection from the normal probe .
note that the inclusion of a normal probe can be interpreted as a source of dephasing@xcite , extended to the case of a supercurrent .
yet it can also have an alternative interpretation . in the presence of superconducting leads ,
superconducting correlations are induced on the dot because of the ( usual ) proximity effect .
this , for instance is illustrated in several recent works both in equilibrium@xcite and in voltage biased@xcite superconducting devices where an effective action for the dot is derived .
now , the addition of a normal lead will induce a reverse proximity effect on the dot which will compete with the existing one : superconducting correlations on the dot will be suppressed .
it is the ratio between the escape rates to the superconducting leads and to the normal leads which will determine the coherent / incoherent character of transport .
the paper is organized as follows . in sec .
ii , the general model is introduced . sec .
iii reviews the non - equilibrium formalism to study transport .
details on how the current harmonics are computed are presented in sec .
the dc current and harmonics are discussed in sec .
v for several limiting cases and in the presence of the normal lead , and the comparison with the josephson current calculation is performed in sec .
we conclude in sec .
vii , while details about the numerical implementation are given in appendix a.
an additional normal metal lead plays the role of a voltage probe@xcite .
its coupling to the dot is controllable and its voltage is taken to satisfy a zero - current condition for the probe.[fig00],width=302 ] the system we study consists of a quantum dot attached to two superconducting leads . for simplicity , the coupling between the two leads is assumed to be symmetric .
in addition , a third lead in the normal metal regime is attached to the dot ( fig .
[ fig00 ] ) .
this model is a generalization of the two terminal structure presented in refs . .
the tunnel coupling from the dot to the normal lead will be a control parameter for adjusting dephasing effects .
the dot contains for simplicity a relevant single electronic level , but the analysis can be straightforwardly extended to treat a multilevel dot .
we label the applied bias voltage on the left ( right ) lead @xmath0 ( @xmath1 ) , where @xmath2 ( @xmath3 ) is measured with respect to the dot level ( see the tunnel hamiltonian below ) .
the chemical potential of the normal lead is set at @xmath4 .
we focus on the most interesting low temperature ( @xmath5 ) limit with moderate ( but not low ) transmission to the superconducting electrodes , which allows in principle to neglect coulomb charging effects .
the resulting hamiltonian of the system with the time - dependent tunnelling terms reads : @xmath6 where @xmath7 the bcs hamiltionian of the superconductors and normal metal is expressed in terms of nambu spinors : @xmath8 with @xmath9 , @xmath10 pauli matrices in nambu space .
the gap is assumed to be the same for the two superconducting leads ( @xmath3 ) @xmath11 , while for the normal lead ( @xmath12 ) @xmath13 . with the nambu notation ,
the tunnelling term becomes : @xmath14 with a tunneling amplitude @xmath15 , where the presence of the bias voltage induces a time dependence of the extracted phases of the superconducting order parameters : @xmath16 , @xmath17 for @xmath18.we often set @xmath19 and restore them when convenient in final expressions .
with these notations , the current operator between lead @xmath20 and the dot reads : @xmath21
we use the keldysh green s function formalism in order to compute the current .
the details of such calculations have appeared elsewhere@xcite , so here we only summarize the techniques .
the coulomb interaction is neglected here .
first , a keldysh partition function is introduced [ @xmath22 : @xmath23 where @xmath24 and @xmath25 @xmath26 is a pauli matrix in keldysh space , and we have introduced a counting field @xmath27 for each lead - dot coupling , which allows to compute the current by deriving the partition with respect to it : @xmath28}{\delta\eta_{j}(t)}\right|_{\eta=0 } ~.\end{aligned}\ ] ] the action possesses a quadratic dependence on the lead fermion spinors , so the latter can be integrated out@xcite .
one is left with an action which depends on the dot spinors only , for which the effect of the leads appears via self energies in keldysh nambu space : @xmath29~ e^{i\sigma_{z}v_{j}t_{2 } } \nonumber \\ \times \ , \left[-\frac{\theta(\delta_{j}-|\omega|)}{\sqrt{\delta_{j}^{2}-\omega^{2}}}\hat{\tau}_{z}+i\,{\rm sign}(\omega)\frac{\theta(|\omega|-\delta_{j})}{\sqrt{\omega^{2}-\delta_{j}^{2}}}\left(\begin{array}{cc } 2f_{\omega}-1 & ~~ -2f_{\omega } \\
+ 2f_{-\omega } & ~~ 2f_{\omega}-1 \\ \end{array}\right)\right],\label{eq05}\end{aligned}\ ] ] where we define the escape rates from the dot as @xmath30 with @xmath31 the ( constant ) density of states of lead @xmath20 in the normal state .
fermi filling factors are given by @xmath32 .
with this definition , the formula for the partial average current reduces to@xcite : @xmath33 where the trace `` @xmath34 '' operates only in nambu space and the symbol @xmath35 denotes convolution in time . here
the green s function of the dot ( which is dressed by the leads ) is defined as : @xmath36 this constitutes an extension of the meir - wingreen@xcite formula for a dot connected to one or several superconducting or normal metal leads .
provided that the green s function is computed exactly it applies also for situations where there are interactions ( electron - electron or electron - phonon ) on the dot .
we are interested in the calculation of the electrical current through a quantum dot with a single level @xmath37 , which is placed near two superconducting leads with applied voltages @xmath38 and @xmath39 .
the reference position for the zero of energy is chosen halfway between the two chemical potentials of the superconductors . in principle , the position of the dot level should depend on the geometry ( escape rates ) and on the applied bias , and it could be derived in a self - consistent way . here , we ignore such dependence and assume that the dot level position can be changed using a metallic gate located close to the dot .
nevertheless , an important part of our study will deal with @xmath40 . in order to avoid the proliferation of parameters
, we choose to specify a symmetric device , where the transparencies of the left and right superconducting leads are the same , while no assumption is made about the normal lead . with this choice , at @xmath41 the requirement that no net current flows in the normal leads imposes that @xmath42 invoking electron / hole symmetry . in
what follows , all energy / frequency scales are expressed in units of the superconducting gap @xmath43 .
when a bias is applied to superconductor , the current is not stationary as it contains all harmonics at the josephson frequency @xmath44 .
it is then convenient to introduce a double fourier transform with summations over discrete domains in frequency : @xmath45 where @xmath46 , and the frequency integration is performed over a finite domain @xmath47 $ ] , where @xmath48 is the voltage on the leads . with these definitions ,
the fourier transform of the current becomes : @xmath49\rf,\label{eq01}\ ] ] where the matrices @xmath50 and @xmath51 have a @xmath52 block structure for every pair of energy domain indexes @xmath53 and @xmath54 . in the double fourier representation ,
the lead self - energy is given by@xcite : @xmath55~,\ ] ] where @xmath56,\ ] ] and @xmath57 .
the dressed green s function for the dot appears from the dyson equation of the form : @xmath58}^{-1}~,\label{inversion_dyson } \label{eq : dressedg}\ ] ] where @xmath59 and @xmath60 .
the dressed green s function is obtained numerically : in practice , it requires the inversion of a `` large '' matrix ( eq . ( [ inversion_dyson ] ) ) .
this finite matrix is obtained by limiting the discrete fourier transforms ( eqs .
( [ eq : doublefourier1 ] ) and ( [ eq : doublefourier2 ] ) ) to a cut - off energy @xmath61 , which has to be large compared to all the relevant energies of the problem .
this energy @xmath61 defines a finite number of frequency domains @xmath62 . as the width of each domain
is @xmath63 , one has @xmath64 , which reflects the fact that at small voltages , one needs to sum on a very large number of andreev reflections . in practice ,
we have chosen in most case @xmath65 , and we have computed the current for @xmath66 only .
see appendix [ sec : appendix ] for more details on the numerical implementation .
we present here the results for the mar currents in different regimes .
first , in sections [ subsec : a ] and [ resonant_section ] , we consider the case without coupling to a normal lead , respectively in the quantum point contact ( qpc ) limit and the the resonant level regime .
this will allow us to compare with existing results , and also to provide new results for the resonant level case .
then , in section [ mar_gamman_section ] , we consider the effect of the coupling to the normal lead on the mar current .
the current is plotted in units of @xmath67 and calculations are performed choosing a temperature @xmath68 in order to study the bias dominated regime . if the escape rate @xmath69 , see eq .
( [ eq : gamma ] ) , is sufficiently large compared to the superconducting gap , the resulting open quantum dot sandwiched between the superconducting electrodes should behave like a point contact .
the case of a point contact was studied both in the context of scattering theory@xcite and from a hamiltonian approach@xcite a decade ago .
there , results were obtained both for the dc current as well as for the first harmonics @xmath70 and @xmath71 of the current . @xmath72
[ figure2 ] displays these quantities for the case of a open quantum dot , for several escape rates and for several level positions of the dot .
the challenge lies with the fact that in order to reproduce the case of the qpc at high transmission ( large coupling @xmath73 ) , one has to include a large number of andreev reflections between the dot and the superconducting leads , especially at low voltages . in practice , this means using a large cut - off energy @xmath61 and thus a very large matrix size @xmath74 for small voltage .
the existing results for the dc current as well as for the first harmonics@xcite provide a reference for our results in the qpc limit . in the top panel of fig .
[ figure2 ] , we plot the dc current for several values of @xmath75 , and several values of the dot level position , which allow to describe the crossover from the high transmission to the low transmission regime . in the normal state ,
the transmission probability between the source and drain electrodes as a function of energy @xmath76 is given by a lorentzian line shape : @xmath77^{-1 } , \label{lorentz}\ ] ] and this transmission is unity ( resonant behavior ) whenever @xmath78 .
this resonant behavior in principle affects the transmission of electrons and holes through the quantum dot . in our calculations to describe the qpc limit , @xmath37 is shifted away from zero beyond the superconducting gap , because otherwise , regardless of the line width broadening provided by @xmath73 , there are reminiscences of resonant mar behavior ( for a description of these resonant processes , see sec .
[ resonant_section ] ) . for @xmath79 and @xmath80 or @xmath81 ,
no significant features of mar are found in the dc current , in accordance with refs . .
the dc current is linear for voltages larger than the quasiparticle onset @xmath82 , and it vanishes abruptly close to @xmath83 , where some mar structure is noticeable for @xmath84 ( lower transparency ) .
note that we have used increased cut - off energies ( @xmath85 and @xmath86 ) for the cases @xmath87 and @xmath88 respectively ; this limits the data for @xmath89 to @xmath90 . in order to further reduce the transparency and
thus observe some structure , one can either decrease the escape rate or drive @xmath37 further away from @xmath91 .
cusps in the current derivative appear as expected at the so - called mar onsets defined as @xmath92 , @xmath53 integer , ( @xmath93 , @xmath94 ) .
as the transparency is reduced ( @xmath93 , @xmath95 and @xmath96 ) only the quasiparticle onset ( @xmath97 ) and the @xmath98 onset are identifiable . by reproducing the calculations of ref .
, we have compared quantitatively our results with those for a true qpc , and found that , by using the transparency @xmath5 of the qpc as a fitting parameter , we can get a near perfect agreement between the two results ( not shown ) .
for example , the values of @xmath5 corresponding to @xmath93 , @xmath99 and @xmath96 are respectively @xmath100 and @xmath101 . in the bottom panels of fig .
[ figure2 ] , the first harmonics are displayed .
we first discuss their general behavior .
both the cosine and sine harmonics display structures at the mar onsets , in the same manner as the dc current .
these structures are more pronounced for lower transparencies ( but note that for @xmath80 , @xmath79 some structure is found for onsets corresponding to @xmath102 ) . for voltages beyond the single particle current ( @xmath97 ) onset , the cosine and sine harmonics eventually decay .
the cosine harmonic vanishes at low voltage , which is clear for all parameters displayed , except for the one which corresponds to the highest transmission ( @xmath89 ) because of our numerical limitation for @xmath103 ( i.e. the vanishing of the cosine harmonics for @xmath89 should be visible for @xmath104 ) .
the sine harmonics saturates at a non zero , positive value , for @xmath105 .
this is a signature of the fact that the josephson effect operates at zero voltage , and our non equilibrium calculation at @xmath103 should in principle match with a calculation of the ( equilibrium ) josephson current .
the value of the first harmonics of the josephson current , computed independently ( see section [ josephson_section ] ) , is shown by a circle on the @xmath106 axis for the low transmission curves .
one can see that there is an excellent agreement between this value and the extrapolation of the non - equilibrium data for @xmath105 ( quantitatively , the agreement is better than 1 % ) . as in the case of the dc current
, we have compared these results with those obtained for a qpc in ref . .
using the values of the qpc transmission obtained with the dc current fits , we find a good qualitative agreement between the two results for the cos component , except at large voltage where the results we obtained decrease ( in absolute value ) much faster . for the sin component ,
the shapes of the curves are similar , but there is an overall downwards shift of the our results with respect to the qpc results . we attribute this to the fact that although the dot level has been positioned away from the superconducting gap region , the energy dependent transmission of the dot still plays a role in the mar processes , and a perfect match with the qpc limit is not attainable .
note that we also find a downwards shift of the first harmonics of the josephson current with respect to the first harmonics of the josephson current of a real qpc@xcite , using the transmission values found with the dc fits .
finally , the fact that the sine harmonic is negative for large @xmath48 ( @xmath107 ) at very large @xmath73 ( @xmath89 or @xmath108 , @xmath79 ) remains a puzzle for us , but it seems to be related to this global downards shift .
we have carefuly checked that this is not due to any numerical imprecision or convergence problem ( see appendix a ) .
the inset on the sine harmonics plot shows the results at large @xmath48 for larger values of @xmath73 ( @xmath109 from bottom to top ) .
it shows that even with larger @xmath73 the sine harmonics remains negative at large @xmath48 , however it tends to decrease in absolute value as @xmath73 is increased .
we now shift the discussion to the case where the line width of the dot is smaller than the superconducting gap , setting for convenience the dot level position at @xmath41 .
the dc current was studied previously for similar parameters@xcite .
here one of our aims is to also include a discussion of the harmonics of the current which is absent in the literature . in the top panel of fig .
[ figure3 ] we show that we are able to reproduce these results for the dc - current .
one finds a very good agreement with existing results : tunnel hamiltonian approach to all orders , or , alternatively scattering theory approach with the lorentzian line shape of eq .
( [ lorentz ] ) . for the dc current ,
the striking effect of the presence of a resonance is the fact that it favors specific andreev reflection processes .
indeed , some structure is found at the odd @xmath53 onsets ( recall that the onsets are located at @xmath92 ) . in the so called mar ladder picture ,
an electron which transits from one superconductor to the other gains an energy @xmath110 , and the same is valid for a hole which is reflected back . because of the presence of the resonance the trajectories of electrons and holes will have larger or smaller weights in the dc current : a resonant trajectory corresponds to the situation where the electron or hole energy crosses at one point the resonant level@xcite .
for the case of @xmath53 odd , with the resonant level located in the middle of the bias window , a resonant hole trajectory always occurs after @xmath111 reflections .
for instance , at @xmath112 the first and last electron trajectory are non resonant , while the hole trajectory is .
we have also observed ( not shown here ) that shifting the position of the resonant level leads to an overall shift of the mar structure . displacing the resonance means that the electron trajectories around the @xmath113 reflection will have an enhanced transmission while the hole resonance weakens . in the top panel of fig .
[ figure3 ] , the dc current for several values of the resonance line width are presented . as expected
, nothing significant occurs for the dc current with decreasing @xmath73 : the mar structure gets sharper , but on the other hand the dc current is reduced . for voltages beyond the superconducting gap , and substantial line widths @xmath114 , the behavior seems to be linear , while for @xmath115 we see the beginning of a saturation for the current at large voltages .
indeed , in the limit @xmath116 the current is only due to quasiparticle transfer and it behaves like the current of a resonant level between two normal leads .
the apparent lack of saturation for @xmath114 only reflects the fact that we are displaying a limited voltage range in order to focus on the mar structure .
we turn now to the first harmonics of the current .
both harmonics decay at voltages beyond the superconducting gap ( quasiparticle dominated regime , as explained above ) .
mar onsets appear at the odd onsets found in the dc current .
however , the cosine harmonic ( fig .
[ figure3 ] lower left panel ) shows a dip at the @xmath98 onset .
it vanishes at low voltage , and displays large voltage oscillations in this range . while the cosine harmonic was strictly negative for the qpc limit , here we see that its sign is reversed , except for the fact that it takes some negative values at relatively high transparencies ( @xmath117 ) near @xmath118 .
the sine harmonic ( fig .
[ figure3 ] lower right panel ) also has a reversed sign with respect to the qpc situation for the voltage range under study , except perhaps at high voltages ( @xmath119 ) and high transparencies .
as the voltage is lowered , the sine harmonic decreases in amplitude and seems to saturate at some negative value when the voltage approaches zero .
one would expect that the sfh would approach the stationary josephson current value when @xmath120 , as it is in a point contact @xcite .
however , this would mean in the present case , that the junction is in the @xmath121-junction regime , which is only possible either under strong coulomb interaction@xcite or in spin active junctions@xcite . in non - interacting non - magnetic resonant contacts
the josephson current is always positive@xcite .
this phenomenon can be explained with strong non - equilibrium population of the andreev levels in the mar regime , which changes their contribution to the current compared to the equilibrium regime . at small bias voltage
, the mar can be understood in terms of adiabatically moving andreev levels and landau - zener transitions between the levels , and between the levels and continuum states@xcite . in point contacts
, the andreev level positions are strongly asymmetric with respect to the chemical potential ( except at the full transmission limit ) : one level lies below the chemical potential close to the filled continuum state band , while another lies above the chemical potential close to the depleted band . at small voltage
the quasiparticles exchange between the levels is negligible , and the level population is determined by an exchange with the corresponding continuum band , thus keeping level populations close to equilibrium . in resonant contacts
the situation is different : when @xmath41 , the levels are symmetric with respect to the chemical potential , they interact with the continuum bands identically , and therefore their populations are identical ( neglecting weak effect of inelastic relaxation ) .
this eventually switches off the level contribution to the josephon current , revealing current of the continuum states , which is negative@xcite . in equilibrium ,
the josephson current is dominated by the andreev levels , and it is positive .
the effect should gradually vanish when the resonant level shifts from the chemical potential , because the andreev level positions become asymmetric as in the point contacts .
we note that similar discontinuity of sfh at zero voltage exists also in fully transparent point contact .
the presence of the resonant level thus modifies drastically the mar current .
it changes the dc current voltage characteristics by favoring some structure at odd mar onsets . for the sine and cosine harmonics ,
the effect is more dramatic as these harmonics change sign in ( almost ) the whole range of the subgap voltage .
we now turn to the central point of this study : how does the gradual coupling to a normal lead affect the mar current ?
first , the dot level is set at @xmath41 , and the chemical potential of the normal lead @xmath122 , in order to insure that no net current flows through this lead .
we focus in this section only on the case of the resonant level regime @xmath123 rather than the point contact regime @xmath124 .
a reason for this choice is first the fact that we expect the dot level to be broadened by the presence of the additional lead , and in order to observe any broadening effect we need to start with a rather sharp level .
second , we have seen in sec . [ resonant_section ] that the presence of a narrow resonance gives rise to rather sharp features in the dc current and its harmonics , so it is natural to ask how these features are modified by the presence of the normal lead . in the top panel of fig .
[ figure4 ] the dc current is plotted , for increasing values of the coupling @xmath125 to the normal lead .
one notices that a moderate amount ( @xmath126 ) of coupling is sufficient to provoke an important deviation with respect to the case in the absence of coupling : at @xmath126 the dc current displays a substantial peak close to @xmath127 , and we are not able to show the decay of this current for @xmath103 because of numerical limitations . in the insets of the panels of fig .
[ figure4 ] , we compare the case @xmath128 and @xmath129 to show that the strong departure from the situation with no coupling is indeed gradual . further increasing the coupling to @xmath130
reduces the peak amplitude , and shifts both this peak and the overall mar structure slightly to the right .
eventually , at @xmath131 , the low voltage peak and the mar onset structure at @xmath132 are lost , the amplitude of the dc current is reduced , but the quasiparticle onset is clearly identifiable . by extrapolation of our data at @xmath103 , we also observe ( for @xmath133 ) that to a very good accuracy the decay of the dc current at low voltage is linear .
however it is known that in the coherent regime , this decay at low voltages is exponential .
this constitutes a signature that the gradual coupling of the dot to the normal lead allows to describe a crossover from coherent to incoherent mar .
the enhancement of the low voltage current for small @xmath134 remains a puzzle .
a similar anomaly occurs at low voltages for the cosine harmonic when @xmath125 is switched on . from the left panel of fig .
[ figure4 ] and its inset , we see a sharp negative value dip at small voltage and small @xmath125 , which becomes more shallow and moves towards larger voltages when the coupling to the normal probe increases . for @xmath135 , the amplitude of the cosine harmonic
is strongly suppressed compared to its value in the absence of normal lead coupling . for @xmath133 , the extrapolation to low voltage of our results
show a linear behavior in voltage .
the presence of the normal lead seems to restore some of the features found for the qpc regime ( negative sign ) because the resonance is broadened .
however , this broadening also erases most of the information on the mar onsets because phase coherence is gradually destroyed . also , note that some dip is formed at @xmath118 , although no noticeable structure was found in the dc current . for the sine harmonic , we saw in sec .
[ resonant_section ] for the resonant case ( with @xmath128 ) that the current seemed to saturate at negative values for @xmath103 . in the presence of coupling to the normal lead however ( fig .
[ figure4 ] , lower right panel ) , this is no longer true : upon reduction of the voltage , this harmonic switches from negative to positive .
the coupling to the normal lead thus restore the equilibrium population of the andreev level for small @xmath48 . in sect .
[ josephson_section ] , we will show that the extrapolation of these curves at @xmath120 truly corresponds to the josephson current .
in contrast to the cosine harmonic , significant mar structures are observed up to @xmath136 , but large @xmath125 eventually sets the sine harmonic to zero . in summary ,
the normal lead acts like a dephasing probe , or , equivalently it is subject to the reverse proximity effect which destroys superconducting correlations on the dot .
first , the coupling to the normal lead tends of course to smoothen the current features , but also it brings a reduction of the supercurrent harmonics because of a lack of phase coherence .
the signatures of mar are due to ( many ) round trips of electrons and holes between the superconducting leads , but if such carriers are brought in contact with a reservoir , only the dc current survives , while the ac current vanishes . for large coupling to the normal reservoir
, the dc current resembles that of two normal metal - superconductor junctions in series , and the linear voltage dependence of the dc current is an illustration of this fact . also , note that the current reduction in the presence of a dephasing lead for a single normal metal - superconductor junction was discussed in ref . .
any resonant feature associated with such a junction gets broadened by the presence of the dephasing probe , and this leads to an effective reduction of the conductance but also of the fano factor .
our microscopic approach of the dc supercurrent and its first harmonics allows to describe the full crossover from coherent transport to incoherent transport .
in this section , we compute to josephson current for the setup including the normal lead .
this allow us to check that the non equilibrium current extrapolated at @xmath120 corresponds to the josephson current.this check was so far performed for a point contact only.@xcite here we extend it to the case of a resonant level , in the presence of the normal lead .
note that this geometry ( josephson junction with a normal lead connected to the central region ) was studied previously @xcite in the context of scattering theory , to probe how the josephson current is affected by current injection from the normal lead .
here we assume no current injection , the normal lead plays a passive role . in order to compute the josephson
current , we use the imaginary - time ( matsubara ) path - integral approach to calculate the partition function @xmath137 and then the josephson current .
the hamiltonian is the same as in sec .
[ model_hamiltonian ] , except that a constant phase difference is imposed between the superconductors .
the lead degrees of freedom are integrated out in the same manner as before .
this yields the partition function ( @xmath138 is a matsubara fermionic frequency ) : @xmath139 with an effective ( euclidean ) lagrangian [ matrix in nambu space ] of the dot @xmath140 where @xmath37 is the dot level measured from the chemical potential of the superconducting leads , @xmath73 and @xmath125 are the tunnel widths of the dot due to its coupling to the left / right superconducting lead and the normal electrode , respectively , @xmath141 is the superconducting phase difference . with these notations , the partition function reduces to a product of @xmath142 determinants : @xmath143 with @xmath144 , and the josephson current is obtained from its logarithmic derivative : @xmath145 note that this expression is similar to that found in ref . , except that the escape rate @xmath125 now appears in the determinant .
the sum is computed numerically and the sine harmonic of @xmath146 is extracted . at the same time , we extrapolate the sine harmonic of the non - equilibrium current ( sec .
[ results_for non_interacting dot_section ] c ) to zero voltage , and we compare the two .
[ figure5 ] shows the comparison , for @xmath123 ( as in section [ mar_gamman_section ] ) .
the extrapolation of the out - of - equilibrium case has been obtained by computing the sine harmonics at @xmath147 and @xmath148 .
we see that the agreement between the sine harmonic of the josephson current and the extrapolated mar harmonics is very good ( there is no adjustable parameter ) , for values of @xmath125 between 0.1 and 2.0 .
note that the precision of the extrapolation increases when @xmath125 increases ( as @xmath71 becomes flatter near @xmath120 when @xmath125 is larger , see fig .
[ figure4 ] ) , which explains that the agreement is not totally perfect for the smaller values of @xmath125 .
going to values of @xmath125 smaller than 0.1(not shown ) leads to some larger discrepancy between the two types of calculations because of the presence of resonant transmission at @xmath37 with unit transmission .
the inset of fig .
[ figure5 ] shows the current phase relationship of the josephson current . for large @xmath125
the current phase relationship becomes essentially sinusoidal , as for a tunnel junction .
the presence of the normal lead randomizes the phase of electrons and holes , and it is equivalent to reducing the transmission probability : we have computed the current phase relationship at @xmath149 , while displacing the dot level @xmath150 $ ] to reduce the transparency of the junction ( not shown ) , and found quantitative agreement between this situation and that where the normal lead is present .
( full curve ) and the extrapolation at zero voltage of the first sine harmonic of the mar current ( circles ) .
inset : current phase relationship of the josephson current for @xmath151 ( from larger to smaller amplitude ) . for both plots
the parameters are @xmath123 and @xmath41 [ figure5],width=302 ]
we have considered non - equilibrium transport through a quantum dot sandwiched between superconducting leads in the subgap regime .
the dc and the first harmonics of the current at the josephson frequency were computed , both in the presence and in the absence of a normal lead connected to the dot . when the dot level is shifted from the middle of the bias voltage window , and the tunneling rates to the superconducting leads
are large compared to the gap , the current resembles closely that of a superconducting qpc .
this is explicit for the dc current , but some difference remain for the harmonics . below the quasiparticle onset , the cosine harmonic clearly identifies with the qpc case , but it deviates from the latter at higher voltages .
the sine harmonic shows qualitative agreement with the qpc situation , except that the overall signal is shifted downwards .
this , for instance , is responsible for a sign change in the sine harmonic at large voltage , while such change is not observed a for a qpc .
this is attributed to the fact that the resonance features of the junction can not be avoided for the parameter chosen , and they affect more the current harmonics than the dc current .
we have also reproduced the dc current of a resonant level in the absence of interactions , and the harmonics of the current have been obtained .
for this situation , we recover the known fact that only the odd mar processes survice the dc current .
these onsets are also present for the harmonics , but the more dramatic effect is that the sign of the current harmonics is reversed with respect to the qpc case . in the presence a coupling to a normal lead ,
the keldysh formalism can be extended by the addition of a self energy in the dot green s function , with appropriate fermi / tunneling phase factors requiring that no current flows in the dot .
so far most of the works on mar have nevertheless focused on either on the coherent or the incoherent mar , and little has been said about the gradual change from one regime to the other ( see although ref . ) .
this has been one of the main focus of this study .
the gradual coupling to the normal lead are at first not dramatic for the dc current .
the coupling to the lead first smoothes out the mar features , but as it is increased a linear ( rather than exponential ) dependence of the current is found at low voltages , and only the quasiparticle onset survives .
this dc current voltage characteristics corresponds to two normal metal - superconductor junctions in series with a total lack of phase coherence between the two .
the harmonics do display a rather dramatic behavior : upon increasing the coupling to the normal lead from zero , the amplitude of both harmonics rises sharply , and it saturates when the tunneling rate to the normal and the superconducting leads are comparable .
the harmonics retain some structure at the predicted onsets for larger coupling than the dc harmonics
. an important check of this work was to compare the first harmonic of the josephson current with the extrapolated value of the mar sine harmonic at zero voltage .
we found that in the presence of the normal lead , the agreement is excellent , but in the absence of so called dephasing ( @xmath128 ) , the resonant features in the mar process render the comparison difficult .
this work could be extended in several directions , some in a straightforward manner : the keldysh formulation which was adopted here is quite flexible and can be adapted to treat multi - terminal devices , or alternatively multilevel dots ( which are likely to give rise to several resonances in the subgap region ) or more complicated structures described by tight binding hamiltonians .
other perspectives of this study concern the inclusion of interactions .
few work include so far electronic correlations on the dot in the calculation of the dc current@xcite .
the approach chosen in ref .
requires a minimal coulomb interaction parameter in order to trigger a departure from the non interacting regime .
there , interactions are shown to reduce drastically the dc - current , to shift and to damp out the structure found at the mar onsets .
the approach of ref .
@xcite is complementary as it treats weak interactions in a systematic perturbation theory approach : there , interactions are shown to lead to a small enhancement of the dc current . at the same time ,
electron - phonon interactions have been included perturbatively @xcite in the calculation of the dc mar current . while inelastic scattering can provide a source of dephasing , the choice of the low temperature regime and the weak electron - phonon coupling assumption resulted there in a current voltage characteristics which is only weakly affected by the vibrations : no dissipation associated with these vibrations was noticed . at higher temperatures and stronger phonon coupling
, phonon scattering could provide a definite source of decoherence . in some sense , the simple model presented here for the inclusion of decoherence could serve as a point of comparison for a more complete study of a dot strongly coupled to phonons out of equilibrium .
the numerical implementation follows directly from equations ( [ eq01 ] ) to ( [ eq : dressedg ] ) .
the matrix inversion giving the dressed green function ( eq . ( [ eq : dressedg ] ) ) can be done once the sum @xmath152 over the frequency domains has been truncated .
we have chosen a constant width @xmath61 for each domain , defining a @xmath48-dependent @xmath153 .
it is clear that the smallest values of @xmath48 are the most expensive to obtain numerically , which explains why our results are limited to @xmath154 .
we have taken great care to check that our results do not suffer from convergence problem due to the truncation . in practice ,
we have chosen @xmath155 , apart from some special cases in the qpc limit where larger @xmath61 were needed ( see the discussion in the qpc section ) .
note that for most of the curves shown in the figures , convergence could be obtained with a much lower value of @xmath61 .
once the dressed green function @xmath156 is computed , the current is obtained by the numerical integration over the fundamental domain @xmath157 $ ] given in eq .
( [ eq01 ] ) .
we have used the following green function sum rules@xcite as independent checks of our method : @xmath158 where @xmath9 acts in nambu space , @xmath159 acts in keldysh space , and the trace is taken on both nambu and keldysh space .
we found that the first sum rule was always satisfied to large precision within our truncation scheme .
the second sum rule is only approximatively satisfied , and gives useful information on the quality of the truncation .
typically , it has a value larger than 1.9 for most of the results shown , but can be somewhat lower when @xmath73 increases .
however , we have found that the computed current does not depend critically on the precise validation of this sum rule ( e.g. in the very large @xmath73 limit , correct value of the current can be obtained even with this sum rule giving values as low as 1.6 ) .
t.m acknowledge financial support from anr pnano grant `` molspintronics '' .
k.b . benefited from the ecole normale - landau institute exchange program .
thanks the centre de physique thorique ( umr 6207 of cnrs ) for its hospitality .
n. van der post , e.t .
peters , i.k .
yanson and j.m .
van ruitenbeek , phys .
lett . , * 73 * , 2611 ( 1994 ) ; e. scheer , p. joyez , d. esteve , c. urbina , * and m. h. devoret , phys .
* 78 * , 3535 ( 1997 ) ; e. scheer , n. agrait , juan carlos cuevas , a. levy - yeyati , b. ludophk , a. martin - rodero , g. rubio bollinger , j. m. van ruitenbeek , c. urbina , nature , * 394 * , 154 ( 1998 ) .
j. kutchinsky , r. taboryski , t. clausen , c. b. srensen , a. kristensen , p. e. lindelof , j. bindslev hansen , c. schelde jacobsen , and j. l. skov , phys .
lett . * 78 * , 931 ( 1997 ) ; t. hoss , c. strunk , t. nussbaumer , r. huber , u. staufer , and c. schnenberger , phys .
b * 62 * , 4079 ( 2000 ) ; c. hoffmann , f. lefloch , m. sanquer , and b. pannetier phys . rev .
b * 70 * , 180503(r ) ( 2004 ) .
e. v. bezuglyi , e. n. bratus , v. s. shumeiko , g. wendin , and h. takayanagi , phys .
b * 62 * , 14439 ( 2000 ) ; j. c. cuevas , j. hammer , j. kopu , j. k. viljas , and m. eschrig , phys . rev .
b * 73 * , 184505 ( 2006 ) ; g. niebler , g. cuniberti , and t. novotny , supercond .
* 22 * , 085016 ( 2009 ) .
m. r. buitelaar , t. nussbaumer , and c. schnenberger , phys
lett . * 89 * , 256801 ( 2002 ) ; m. r. buitelaar , w. belzig , t. nussbaumer , b. babic , c. bruder , and c. schnenberger , phys .
lett . * 91 * , 057005 ( 2003 ) .
h. i. jorgensen,1 , k. grove - rasmussen , t. novotny , k. flensberg , and p. e. lindelof phys .
* 96 * , 207003 ( 2006 ) ; t. sand - jespersen , j. paaske , b. m. andersen , k. grove - rasmussen , h. i. jrgensen , m. aagesen , c. b. srensen , p. e. lindelof , k. flensberg , and j. nygrd , phys . rev . lett . * 99 * , 126603 ( 2007 ) . li - fu chang and philip f. bagwell , phys .
b * 55 * , 12678 ( 1997 ) ; p. samuelsson , v.s .
shumeiko , and g. wendin , phys .
b , * 56 * , r5763 ( 1997);p .
samuelsson , j. lantz , v.s .
shumeiko , and g. wendin , phys .
b * 62 * , 1319 ( 2000 ) .
|
we study the out - of - equilibrium current through a quantum dot which is placed between two superconducting leads held at fixed voltage bias , considering both cases of the absence and the presence of an additional normal lead connected to the dot . using the non - equilibrium keldysh technique
, we focus on the subgap bias region , where multiple andreev reflections ( mar ) are responsible for charge transfer through the dot . attention is put on the dc current and on the first harmonics of the supercurrent .
varying the position and/or the width of the dot level , we first investigate the crossover between a quantum dot and quantum point contact regimes in the absence of a normal lead .
we then study the effect of the normal electrode connected to the dot , which is understood to lead to dephasing , or alternatively to induce reverse proximity effect . by increasing the dot coupling to the normal probe
, we show the full crossover from zero dephasing to the incoherent case .
we also compute the josephson current in the presence of the normal lead , and find it in excellent agreement with the values of the non - equlibrium current extrapolated at zero voltage .
|
classification of matter into different ordered phases is a central theme in condensed matter physics .
traditionally , ordered phases have been understood using the ginzburg - landau paradigm,@xcite where spontaneous symmetry breaking gives rise to a finite order parameter , such as a finite magnetization in a ferromagnet .
starting with the integer quantum hall effect,@xcite another classification paradigm based on the topological properties of the band structure has also emerged.@xcite with the recent discovery of topological insulators it has now become a reoccurring task to classify different phases of matter according to the topological properties of their ( single - particle ) band structure
. superconductors with a full pairing gap can be classified into different topological classes in much the same way as band insulators.@xcite non - trivial topological superconductors have lately received an enormous amount of attention,@xcite due to the possibility of them hosting majorana fermions at e.g. interfaces and vortices.@xcite the majorana fermion is its own anti - particle and obeys non - abelian statistics , which can be used for fault - tolerant topological quantum computation.@xcite the majorana state is a consequence of the bulk - boundary correspondence , which establishes that non - trivial topological order in the bulk necessarily gives rise to boundary states crossing the bulk energy gap.@xcite in some topological superconductors the resulting zero energy modes are majorana fermions . proposals for 2d topological superconductors with majorana fermions exist already , in e.g. superconducting hybrid structures between a conventional @xmath0-wave superconductor and topological insulators or ferromagnetic and rashba spin - orbit coupled semiconductors with finite magnetization.@xcite the latter alternative is particularly interesting since semiconductor technology is very mature , there exists semiconductors with large spin - orbit coupling@xcite and with demonstrated superconducting proximity effect@xcite , while ferromagnetism can be introduced either through proximity to a ferromagnetic insulator or through an applied magnetic field .
all key ingredients are thus in place for finding majorana fermions.@xcite in fact , possible experimental signatures of majorana fermions have already been reported in similar one - dimensional ( 1d ) nanowire semiconductor structures.@xcite although topological classification is a very powerful concept , the topological invariant usually remains a rather abstract quantity .
the most prominent example is probably the chern number , used for 2d time - reversal symmetry breaking systems,@xcite including the 2d ferromagnetic semiconductor - superconductor heterostructures mentioned above.@xcite in this article we show that the topological phase for these heterostructures instead can be classified by a sum of the skyrmion number for the spin polarization in each band ; that is , according to how the spin texture of the different bands wraps around the unit sphere , with the topological invariant given by the sum of the topological spin texture in each band .
the skyrmion classification gives simultaneously both the topological invariant and a very intuitive physical picture of its meaning , thus making the topological order more accessible .
most importantly , the spin skyrmion numbers can be directly measured using a spin - polarized band structure imaging technique such as spin - polarized angle - resolved photoemission spectroscopy ( arpes ) .
this opens up for direct experimental classification of the topological phase without having to rely on indirect methods such as the existence of boundary modes .
moreover , we show why such imaging need only to be performed at certain high symmetry points , which is an important practical advantage .
we also extend the results for the 2d case to 1d wires@xcite for which experiments has reported majorana fermion signatures.@xcite we do so through dimensional reduction of the 2d case , and show that the spins covers a circle rather than a full sphere in the topologically non - trivial phase .
this provides a unified framework within which both 1d and 2d systems can be studied .
ferromagnetic semiconductor - superconductor heterostructures with strong rashba spin - orbit interaction in 2d have previously been topologically classified using a chern number.@xcite in general , the chern number for a hamiltonian @xmath1 is given by@xcite @xmath2 where eq .
( [ equation : chern_number ] ) , ( [ equation : berry_curvature ] ) and ( [ equation : berry_connection ] ) are the chern number , berry curvature , and berry connection , respectively . here
@xmath3 is the eigenstate of @xmath1 in band @xmath4 at point @xmath5 in the brillouin zone .
it is the sum of the chern numbers for all occupied bands that is used in actual classifications , but we will here consider the chern number for each band individually and only add them in the end .
this is in general allowed as long as separate bands , with physical properties continuous in @xmath5 , can be unambiguously defined .
one exception is the topological phase transition , where the band assignment is necessarily ambiguous . here
we will show that the chern number defined in eq .
( [ equation : chern_number ] ) for the particular case of a two - component spinor field over a 2d compact manifold can be viewed as a skyrmion number .
for all practical purposes the compact space is the brillouin zone and the skyrmion number simply tells us how many times the spin covers the unit sphere as it sweeps the brillouin zone .
this is a previously known result@xcite but expressed using a slightly different formulation that prepares for the results in the sections that follows .
we thus assume a single two - spinor field @xmath6^t$ ] defined over the brillouin zone , where @xmath7 is the up ( down ) spin index , and an arbitrary phase factor @xmath8 has been factored out .
the berry connection in eq .
( [ equation : berry_connection ] ) is then given by @xmath9 the third term disappears when calculating the berry curvature in eq .
( [ equation : berry_curvature ] ) and will therefore be ignored .
this allows us to redefine @xmath10 to be the sum of the two first terms only , while at the same time assuming that @xmath11 is real by letting its phase dependence be captured by @xmath8 .
an exception to this construction occurs at points where @xmath12 , because at these points a unique @xmath13 can not be defined .
however , at these points the berry connection is given by an expression of the form @xmath14 where @xmath15 is the full complex phase of the down component .
the first term in this expression gives zero contribution because it is real , while the second term cancels out in the same way as the phase factor in eq .
( [ equation : berry_connection_two_spinor ] ) canceled out when the berry connection in eq .
( [ equation : berry_curvature ] ) is calculated .
we therefore see that the contribution to the chern number from such points are zero , and we simply chose do redefine the integration region in eq .
( [ equation : chern_number ] ) such that these points are excluded from the integral .
having justified the assumption of a real @xmath11 , we now define the three quantities @xmath16 where @xmath17 and @xmath18 are the real and imaginary parts of @xmath19 .
together these three expressions defines a unit vector @xmath20 , which at each @xmath5 in the brillouin zone points in the spin direction . inverting these expressions and using them in eq .
( [ equation : berry_connection_two_spinor ] ) , the berry connection becomes @xmath21 this expression contains an apparent singularity at @xmath22 , but this happens exactly at those points for which @xmath23 , points that we have already excluded from the integration domain .
these apparent singularities do therefore not present any problem .
it follows from the expression in eq .
( [ equation : berry_connection_four_spinor ] ) that the berry curvature is given by @xmath24 where @xmath25 .
now , @xmath26 can be decomposed as @xmath27 , where @xmath28 is perpendicular to @xmath20 and therefore also to @xmath29 , since @xmath20 is a unit vector .
thus the chern number obtained from eq .
( [ equation : chern_number ] ) takes the form @xmath30 we note that the integrand in this expression gives zero contribution in the regions where @xmath22 , because either the cross product is zero , or the region on which @xmath22 has zero measure .
it is therefore safe to re - extend the integration back to the full brillouin zone .
the expression in eq .
( [ equation : skyrmion_number ] ) is nothing but the winding number for a vector taking values on the unit sphere @xmath31 , or because the unit vector represents a spin direction , a spin skyrmion number .
we therefore conclude that when @xmath3 is a two - component spinor field defined over the 2d brillouin zone , the chern number is a skyrmion number .
now we turn our attention to the chern number that is used for the classification of 2d semiconductor heterostructures which hosts majorana fermions in the topologically non - trivial phase.@xcite more precisely , we consider here the chern number for four - component spinor fields which arise as solutions to the eigenvalue equation for hamiltonians of the form @xmath32,\ ] ] in two spatial dimensions , which is written in the basis @xmath33^t = \left[\begin{array}{cccc}u_{\uparrow\mathbf{k}}^{(\lambda ) } & u_{\mathbf{\downarrow\mathbf{k}}}^ { ( \lambda ) } & v_{\uparrow-\mathbf{k}}^ { ( \lambda ) } & v_{\downarrow-\mathbf{k}}^ { ( \lambda)}\end{array}\right]^t$ ] . here
@xmath34 and @xmath35 are the electron and hole components of the eigenstates .
( [ equation : hamiltonian ] ) is the most general bogoliubov - de gennes hamiltonian , and a ferromagnetic semiconductor - superconductor heterostructure can be modelled by setting@xcite @xmath36 here @xmath37 is the band structure of a lightly doped semiconductor which we approximate by @xmath38 with @xmath39 , which is the simplest possible band structure on a square lattice .
furthermore , @xmath40 provides a rashba spin - orbit coupling and @xmath41 is a zeeman exchange field .
finally @xmath42 provides an @xmath0-wave superconducting term induced through proximity effect to a conventional superconductor .
because the combination of spin - orbit interaction and zeeman term forces the bogoliubov - de gennes hamiltonian to be written on a four - by - four form , the degrees of freedom has been artificially doubled .
we therefore expect the information of the system , such as the topological invariant , to be redundantly encoded in the resulting spinors .
the purpose here is to show that it is possible to use only half of this information to calculate the topological invariant .
especially we will show that this construction leads us to an interpretation of also this chern number as a skyrmion number .
the first step in this process lies in realizing that because the order parameter is nowhere zero , @xmath42 can be inverted , which together with the two first lines of the eigenvalue equation @xmath43 results in @xmath44 this expression , together with the normalization condition @xmath45 , implies that @xmath34 and @xmath46 is always finite .
the importance of this equation is that it shows that the electron components alone contains all the information about the system .
further it allows us to define @xmath47 the renormalized spin component .
it follows from the discussion in the previous section that , because @xmath48 is a normalized two - component spinor , it is possible to define a skyrmion number @xmath49 this is clearly a topological invariant , which can only change when two bands cross each other . to different states becomes arbitrary at the degenerate points . ]
we now show that the skyrmion number in eq .
( [ equation : skyrmion_number_four_component ] ) is the same as the chern number calculated from eq .
( [ equation : chern_number]-[equation : berry_connection ] ) for the full four - component spinor . for this purpose
we plot , for multiple parameter values @xmath41 and @xmath42 , both the chern number and skyrmion number side by side for all four bands in fig .
[ figure : skyrmion_vs_chern ] .
this is done by numerically diagonalizing eq .
( [ equation : hamiltonian ] ) and using eq .
( [ equation : skyrmion_number_four_component ] ) to calculate the skyrmion number .
the chern number is in principle given by eq .
( [ equation : chern_number]-[equation : berry_connection ] ) , but the well known alternative expression@xcite @xmath50 has instead been used for numerical reasons .
[ figure : skyrmion_vs_chern ] shows that the two expressions clearly agree for all plotted values . and
@xmath42 for a 2d ferromagnetic and rashba spin - orbit coupled semiconductor - superconductor heterostructure .
band 1 and 4 are topologically trivial for all @xmath41 and @xmath42 , while band 2 and 3 acquire a topologically non - trivial configuration for @xmath51 .
x and o mark the values for which spin configurations are shown in fig .
[ figure : spin_configurations ] . ]
one might here object that the two expressions have only been shown to agree for certain parameters .
however , we already know that both expressions are topological invariants , which can only change when bands cross each other .
if they agree at a single point in one topological phase , they must therefore agree at all points in the same phase .
the only possibility to find another set of parameters where they do not agree would be if the system passed through a topological phase transition and the change in topological number was different across the transition for the two expressions .
however , we clearly see that they agree on both sides of the topological phase transition in fig [ figure : skyrmion_vs_chern ] .
further , this is the only phase transition of interest in realistic materials involving semiconductors,@xcite we therefore conclude that it is safe to use the skyrmion expression instead of the full four component chern number .
the main advantage of the skyrmion number formulation over the more abstract chern number is that eq .
( [ equation : skyrmion_number_four_component ] ) has a direct physical significance : it is the winding of the spin direction for the electrons at different points in the brilloiun zone .
this allows for the topological invariant to be directly extracted using any method where spin - resolved band structure data is available .
in fact , eq .
( [ equation : skyrmion_number_four_component ] ) itself becomes nearly obsolete once spin - polarized band structure data is obtained , as the topological invariant can be deduced from simple inspection , as seen in fig .
[ figure : spin_configurations ] . here
we have plotted the spin configurations both in the trivial and non - trivial phase for the specific parameter values marked by @xmath52 and @xmath53 in fig .
[ figure : skyrmion_vs_chern ] . as clearly visible
, skyrmions appears in the two bands closest to the band crossing .
we here also mention that eq .
( [ equation : berry_connection ] ) is antisymmetric with respect to the particle - hole symmetry , and the skyrmion numbers in the particle - hole related bands are therefore also related by a minus sign .
the formalism developed above for 2d semiconductor heterostructures can be expanded to 1d wires .
this is particularly interesting as the so far most promising system for realization of majorana fermions are 1d wires consisting of a similar combination of ferromagnet - semiconductor - superconductor components .
this system can be described by using the hamiltonian @xmath54 in eq .
( [ equation : hamiltonian ] ) , and is up to choice of @xmath55-matrix labels the lattice equivalent of previously studied continuum models.@xcite we arrive at this expression through dimensional reduction of eq .
( [ equation : hamiltonian_specific ] ) by setting @xmath56 and @xmath57 .
thus the spin structure of the 1d model follows from that along @xmath57 in fig .
[ figure : spin_configurations ] .
in addition to the 1d case being of large interest in itself , a detailed study of it also gives a better understanding of the origin of the skyrmions in 2d . in fig .
[ figure : one_dimensional_band_structure ] we plot the band structure for the 1d wire in eq .
( [ equation : one_dimensional ] ) for representative values of the different topological phases , as well as the band structure for @xmath58 . on top of these bands
the renormalized spins given by eq .
( [ equation : renormalized_spin ] ) are also indicated . for @xmath58
all spins in any specific band is either parallel or anti - parallel with the zeeman term @xmath41 . as @xmath41 is turned up , the @xmath58 bands with up spins are pushed down in energy , while the down spin bands are pushed up . in the intermediate parameter region ( b ) , two bands with opposite spin overlap each other .
because these two bands bend in opposite directions , a situation is created where at some points in the brillouin zone down spins are below the fermi level , while at other points instead up spins are below .
once @xmath59 is turned on , the spin structure does not change at @xmath60 where the spin - orbit interaction has no effect .
most importantly , in the topologically non - trivial region ( b ) , the degeneracy between the two @xmath58 bands at the crossing points at the fermi level is lifted .
this results in two separate bands around the fermi level , which have different spin direction at @xmath61 and @xmath62 .
away from these high symmetry points the @xmath59 term twists the spin into the @xmath63-direction , thereby creating a smooth transition from the up(down ) spin at @xmath61 , to the down(up ) spin at @xmath62 in the upper(lower ) band .
finally imagine turning up @xmath41 even further before turning on @xmath59 .
although self - consistent calculations has shown that this is a rather unphysical situation , since @xmath41 destroys superconductivity,@xcite this case is useful to consider for conceptual purposes .
eventually we reach the situation where the @xmath58 bands no longer overlap , such that all up spins are below the fermi level while all down spins are above .
this is represented by the band structure in fig .
[ figure : one_dimensional_band_structure](c ) . in this case
the spins at @xmath61 and @xmath62 are the same in each band and the only effect of turning on @xmath59 is to somewhat deform the spin structure .
this deformation no longer twists the spin around a full circle as we move along @xmath64 in @xmath65-space , but rather just twists it toward the @xmath63-axis and then back toward the same @xmath66-direction again .
it is in this sense most useful to think of the topologically non - trivial phase as the parameter interval where the spin at one , but only one , of the points @xmath61 and @xmath67 has been interchanged between the two bands closest to the fermi energy , thereby forcing the spin to twist from up to down and back to up again as we trace out the full 1d brillouin zone .
this is the very origin of the twisted spin structure . in fig .
[ figure : one_dimensional ] we provide further information which link the spin structures , band crossings , and existence of majorana fermions to each other . in the energy spectrum ( top )
we clearly see that the region between ( @xmath68 ) and ( @xmath69 ) hosts the characteristic majorana zero mode . comparing to the band structures in fig .
[ figure : one_dimensional_band_structure ] , the band structure in [ figure : one_dimensional_band_structure](a ) transforms into the band structure [ figure : one_dimensional_band_structure](b ) at the point which is marked by ( @xmath68 ) in fig .
[ figure : one_dimensional ] , while the transition from the band structure in [ figure : one_dimensional_band_structure](b ) to that in [ figure : one_dimensional_band_structure](c ) happens at ( @xmath69 ) .
in other words , ( @xmath68 ) is the point where the two bands closest to the fermi level meet at @xmath62 , while ( @xmath69 ) is the point where they meet at @xmath61 .
thus the majorana fermion appears in the intermediate region where one , but not both , of the spins at the high symmetry points has interchanged band with each other . in fig .
[ figure : one_dimensional](bottom ) we also clearly see that the spin structure forms a circle in this and only this intermediate region ( shaded red ) . for a 1d wire with open boundary conditions .
topologically non - trivial region @xmath70 $ ] hosts majorana fermions , seen as horizontal lines at @xmath71 .
interchange of spins between the bands at @xmath62 and @xmath61 , marking the entrance / exit of the topologically non - trivial phase , occurs at a and b. ( bottom ) spin structure in the four bands for the same three representative values of @xmath41 as in fig .
[ figure : one_dimensional_band_structure ] , and as indicated by the vertical lines at the top . inside the non - trivial phase the spin structure in the two bands closest to the fermi energy covers a whole circle ( shaded red ) . ] from these observations it is clear how the one - dimensional analogue of eq . ( [ equation : skyrmion_number_four_component ] ) is to be defined , namely @xmath72 in the case a different combination of @xmath55-matrices are used in the hamiltonian in eq .
( [ equation : one_dimensional ] ) , the unit vector @xmath73 should be replaced by the unit vector which corresponds to the @xmath55-matrix which is not used in the hamiltonian . to understand why this is the correct expression for the topological invariant , it is enough to realize that the expression inside the parenthesis is a vector pointing in the @xmath73-direction , with magnitude equal to the rate of change of @xmath20 .
also note that although the expression in principle is able to calculate an arbitrary winding of @xmath74 , in reality @xmath20 will only give rise to @xmath75 , because the spin orbit interaction is only able to twist the spin once across the 1d brillouin zone .
the sign is related to whether the spins are twisted clockwise or counter - clockwise , and is changed if @xmath76 or @xmath77 , but is of no importance as far as the existence of majorana fermions is considered .
that is , the topological invariant is effectively a @xmath78 invariant . above we showed how the topologically non - trivial spin structure arises in the 1d case when the spins at one , but not both , of the high symmetry points @xmath60 interchange band with each other . therefore making the spin at @xmath61 and @xmath62 within the same band point in opposite directions .
it is now worthwhile to revisit the 2d case
. a quick look at fig .
[ figure : spin_configurations ] reveals that the observed skyrmion arise in a similar way ; it arise when the spin at different high symmetry points in the same band point in different directions .
also in this case it is the spin - orbit interaction which provides the smooth transition between these conflicting spin directions in the region between the high symmetry points , thereby creating a skyrmion .
this is a very powerful observation from an experimental point of view , because this allows for identification of the topological phase to be performed by mapping out the spin direction only at the high symmetry points .
if the system has a full energy gap , and any spin at a high symmetry point not agrees with the spin direction at another high symmetry point in the same band , then a skyrmion necessarily have developed in that band .
in summary we have shown that the topological phase of 2d ferromagnetic semiconductor - superconductor heterostructures , previously classified using chern numbers , and their 1d analogues , most easily can be thought of as phases hosting band structure skyrmions .
the skyrmion topological invariant is the sum of the number of momentum - space skyrmions in the spin configuration in the occupied bands .
this result shows that any technique capable of extracting the spin - resolved band structure can be used to determine the topological phase of the system .
moreover , it is enough to map out the spin direction only at the high symmetry points , if in addition the system is known to be gapped throughout the brillouin zone . in the case
some spin direction at some high symmetry point do not agree with the spin direction at another high symmetry point in the same band , then a skyrmion has developed in the band , and the system is in a topologically non - trivial phase .
thus for example arpes seems ideally suited to directly measure the topological invariant .
indicative information is also available in the topologically trivial phase by studying the spreading of the spins in the full brillouin zone .
if the spins are observed to cover an ever greater area of the unit sphere as the bands approach each other towards a band crossing across the energy gap , then that is a clear indication of the band crossing being a topological phase transition .
this could prove especially useful in cases where different band structures are engineered but a potential topological phase transition is initially difficult to obtain .
our treatment of the 1d case also makes it clear that the 1d wires and 2d heterostructures can be treated within the same framework .
the only difference between the 1d and 2d case is that the spins in the 1d case traces out a circle instead of a full skyrmion .
we think that this unified framework , and the intuitively more accessible skyrmion number , will prove helpful in both experimental investigation of these materials , as well as in theoretical model building .
due to the physical accessibility of the spin configuration by a spin - resolved band structure imaging techniques , be it experimental or numerical , this result should significantly accelerate the search for topologically non - trivial superconducting states .
we are grateful to j. c. budich for discussions and the swedish research council ( vr ) for financial support .
y. l. chen , j. g. analytis , j. h. chu , z. k. liu , s. k. mo , x. l. qi , h. j. zhang , d. h. lu , x. dai , z. fang , s. c. zhang , i. r. fisher , z. hussain , and z. x. shen http://dx.doi.org/10.1126/science.1173034[science * 325 * , 178 ( 2009 ) . ]
m. sakano , m. s. bahramy , a. katayama , t. shimojima , h. murakawa , y. kaneko , w. malaeb , s. shin , k. ono , h. kumigashira , r. arita , n. nagaosa , h. y. hwang , y. tokura , and k. ishizaka , http://dx.doi.org/10.1103/physrevlett.110.107204[phys .
110 * , 107204 ( 2013 ) . ]
|
we provide a derivation of a spin skyrmion number classification for two - dimensional topological superconductors constructed from ferromagnetic and rashba spin - orbit coupled semiconductor - superconductor heterostructures .
we show that in the non - trivial topological phase , characterized by a non - zero chern number , there is always a topological spin texture in the occupied bands represented by a skyrmion number .
the skyrmion number has the advantage of being both physically intuitive and directly measurable using spin - sensitive band structure imaging techniques .
in addition , we show that the skyrmion classification can be extended to the equivalent one - dimensional topological superconductors .
|
defects usually play a prominent role in various properties of a solid . for this reason , the physics and chemistry of defects have been the subjects of much study for several decades.@xcite many of these works focused mainly at the dilute limit , _
i.e. _ , with a small defect concentration .
this is the case of interest in doped semiconductors and/or compounds in the immediate vicinity of the stoichiometry . at certain conditions ( usually with high temperature ) deviation from the stoichiometry
can span over a wide range of chemical composition .
binary oxides ceo@xmath0 and uo@xmath0 represent the paradigms for such kind of non - stoichiometry in the fluorite - related structures , and many others exist as well.@xcite for a comprehensive understanding of these materials , knowledge from just near the stoichiometry is insufficient .
this is because many physical quantities depend strongly on non - stoichiometry , and exhibit quite different behavior when at finite deviations . on the other hand , defects at high pressure
have received little attention so far , and our knowledge about their general behavior at highly compressional conditions is very limited , in spite of the fact that they are crucial for earth modeling and for planetary evolution description , where plenty of defects presenting in variety of non - stoichiometric minerals in the interior of these celestial bodies .
the capability to capture correctly the energetics and other physical properties across the whole stoichiometry range at different temperature - pressure conditions is an essential requirement for the purpose of predicting and controlling the behavior of these complex materials .
nevertheless , a general theoretical method for this purpose is still elusive .
even at ambient conditions and near the stoichiometry , our understanding about nuclear oxides such as uo@xmath0 and ( u , pu)o@xmath0 is also limited and unsatisfactory.@xcite although there are many papers and reports have been published on various aspects of diffusion in these oxides , a reasonable level of understanding has been reached only in the case of oxygen , from which the formation energy of oxygen frenkel pair can be deduced when the migration energy of the corresponding diffusion process is known.@xcite nevertheless , measuring oxygen diffusion in stoichiometric oxides is difficult because of the need to maintain the stoichiometry , which is almost impossible for a large temperature range.@xcite the difficulty also lies in the interpretation of available experimental data . while theoretical calculations can be applied to individual processes , transport and other data often correspond to a superposition of several entangled processes , and makes extraction of the desired information complicated.@xcite in addition to chemical diffusion and self - diffusion , electrical conductivity@xcite and neutron scattering@xcite have also been performed to measure the defect concentrations and formation energies . in all of these experiments , a poor understanding of the experimental conditions , as well as the inherent difficulties in the measurement and subsequent interpretation of data , have caused dispersed results.@xcite there are also very few experimental data existed for cation defects in nuclear oxides.@xcite for example in uo@xmath0 , only crude estimates of the activation energy for uranium self - diffusion and migration energy of uranium vacancy are available . with the aid of point defect model ( pdm ) , one can extract the schottky defect formation energy from these estimates .
but since a great uncertainty remains in the experimental data , the reliability of the derived value is doubtful.@xcite progresses in density functional theory of electrons and in computational algorithms make it possible to calculate the relevant energetics directly from quantum mechanics .
such kind of first principles methods have provided better data than previous semi - empirical interatomic potentials , and are comparable to experimental measurements .
the electronic structures,@xcite structural phase transformations@xcite and equations of state ( eos),@xcite oxygen diffusion,@xcite and some defect clustering structures@xcite have been modeled .
unfortunately , this advance never reached a satisfactory level for defect energetics , even though a lot of efforts have been devoted to it.@xcite it even can not reproduce the experimental fact that oxygen defects dominate the whole stoichiometry range.@xcite also one should note that including oxygen clusters can give rise to the expected predominance of oxygen defects,@xcite but its relevance to the experimentally measured defect formation energy is unclear.@xcite this discrepancy among experimental and theoretical results might be due to the limited accuracy of the theoretical assessments , for example the possible meta - stable electronic states that could be encountered in calculations of strongly correlated materials,@xcite the approximation employed to treat the partially localized 5__f _ _ orbitals,@xcite the variation of the charge state of defects,@xcite and the small size of the simulation cell for defective structure modeling that currently accessible , _ etc . _
all of these might render uncertain error in the final results .
but it also can arise from the error lying in the experimental estimates that widely used as the benchmark for theoretical modeling : the data are scattering and not fully self - consistent , and in some cases these estimated data can not reproduce the fact of the predominance of oxygen defects too.@xcite this makes the problem entangled and very hard to treat with .
therefore any approach that can constrain the defect energetics and reduce its uncertainty is of decisive help for solving the problem . in this report
, we investigate the possibility of a such kind of method by examining the physics that governs the pressure - driven pseudo - transitions@xcite between different defect species . by establishing a theoretical relationship of the pseudo - phase boundary on defect formation energies ,
we show that strict constraints can be imposed on these energy parameters , which might then be used to refine the experimental estimates . to achieve the final goal , however , one has first to measure the curve of pseudo - transition pressures , and obtain accurate equations of state of defects , then by making use of the theoretical method that we will present below , to get reliable estimate of defect formation energies . in this sense
this report is the first step but also the most important step towards this accomplishment . by developing the theoretical basis of this constraining procedure , it not only provides us a new angle to understand the long - standing problem , but also establishes a general method in treating highly defective materials under high pressures . for the clarity of discussion
, we will first present a simple theoretical framework using pdm that is intuitive and easy to understand , then a generalization to the general case will be made . in sec.[sec : simple_model ] we discuss the pdm that allows us to calculate the pseudo - phase diagram of point defects , and then establish a relationship between pseudo - phase boundary and defect energetics . using this powerful tool ,
the influence of intrinsic defect formation energy on pseudo - transition is then investigated in sec.[sec : pdm ] .
the obtained information represents a constraint on the possible value of the formation energy of the defects .
though our discussions are mainly focused on uranium dioxide in this paper , we also extend the investigation by considering virtual models to explore other interesting phenomena such as reentrant pseudo - transition and multi - defect coexistence in sec.[sec : virtual system ] .
it is well known that pdm does not take defect - defect interactions into account and can be applied to only the vicinity of the stoichiometry . in order to deal with highly defective region and extrapolate the energetic and thermodynamic information obtained at finite stoichiometry deviations to the dilute limit , a general formalism that can treat defect interactions is developed in sec.[sec : defect interactions ] .
this generalization is necessary for a realistic description of the nonlinear dependence of thermodynamic properties on non - stoichiometry . a brief discussion and remarks on charge state of defects , as well as other relevant issues ,
are then given in sec.[sec : discussion ] , which is followed by a summary of the main conclusions .
at dilute limit where defect concentration is negligibly small , imperfectness in crystal has little impact on thermodynamic and mechanical properties , despite a profound modification on electrical conductivity and/or magnetism might often occur . at large stoichiometry deviation ,
the defect concentration is governed by the ratio of chemical compositions rather than by thermal excitation , therefore the density of defects could be enormous . in this case , noticeable influence on general thermodynamic quantities can be expected , so as on relevant mechanical properties . to understand the general effects of non - stoichiometry on material s behavior , we require a physical model that expresses the defect density as a function of external conditions usually the hydrostatic pressure and temperature , and how the presentation of defects modifies thermodynamic functions such as enthalpy or gibbs free energy .
having such a function that incorporated defect effects , all relevant thermodynamic properties can be derived straightforwardly . in this section
we first present the basic picture by considering the simple pdm .
a general formalism will be developed in sec.[sec : defect interactions ] . in pdm,@xcite the spatial size of an individual defect
is assumed to be of zero dimension , and all interactions among them are neglected . in this simple model , defect concentrations
are determined by the corresponding formation energy of isolated defects . considering a structure that contains one defect of type @xmath1
, its gibbs free energy can be written as @xmath2 in which @xmath3 , @xmath4 , and @xmath5 stand for hydrostatic pressure , temperature , and volume of the simulation cell , respectively .
the cohesive energy at zero kelvin in static approximation reads @xmath6^{2 } \label{eq : colde}\ ] ] when expressed in birch - murnaghan equation ( other eos model can be used as well ) . here
variables with subscript 0 denote the corresponding value in the equilibrium condition of zero pressure , and @xmath7 is the bulk modulus .
the contribution of lattice dynamics to the free energy can be approximated in the debye model as @xmath8-k_{b}tf\left(\theta_{d}/t\right)+\frac{9}{8}k_{b}\theta_{d},\ ] ] where @xmath9 is the boltzmann constant and @xmath10 the debye function .
the debye temperature can be evaluated approximately by@xcite @xmath11^{1/2}.\ ] ] here the superscript @xmath12 denotes the reference state ( here the defect - free uo@xmath13 ) , @xmath14 is the effective volume per atom , and @xmath15 is the effective atomic weight .
the parameters in these equations , namely @xmath16 , @xmath17 , and @xmath18 , can be obtained by fitting to _ ab initio _ results of density functional theory , while @xmath19 can be taken from x - ray diffraction measurement.@xcite the details of determining the value of these parameters have been discussed and presented in ref ..
having known @xmath20 , the formation gibbs free energy ( fge ) of intrinsic point defects can be constructed .
for example , the fge of a frenkel pair ( fp ) of species @xmath21 is then expressed as @xmath22 and for the schottky defect ( s ) as ( taking uo@xmath0 as the example ) @xmath23 here @xmath24 denotes the number of atoms in a defect - free cell and @xmath25 is the corresponding gibbs free energy , @xmath26 is the gibbs free energy of the cell containing the respective defect . in a closed regime where no particle exchange with the exterior can occur ,
the defect concentration must satisfy@xcite @xmath27[i_{o}]=\exp\left(\frac{-\delta g_{\mathrm{o{\_}fp}}}{k_{b } t}\right)\label{eq : pdm1},\\ & [ v_{u}][i_{u}]=\exp\left(\frac{-\delta g_{\mathrm{u{\_}fp}}}{k_{b } t}\right)\label{eq : pdm2},\\ & [ v_{o}]^{2}[v_{u}]=\exp\left(\frac{-\delta g_{\mathrm{s}}}{k_{b } t}\right)\label{eq : pdm3}.\end{aligned}\ ] ] the composition equation that expressed in point defect populations is @xmath28-[i_{u}]\right)+[i_{o}]-2[v_{o}]}{1-[v_{u}]+[i_{u } ] } , \label{eq : pdmx}\ ] ] where @xmath29 is the stoichiometry deviation ( for example that in uo@xmath30 ) .
notice that eq.([eq : pdmx ] ) is different from the conventional definition of @xmath31-[i_{u}]\right)+[i_{o}]-2[v_{o } ] , \label{eq : pdm4}\ ] ] which is valid only when no cation defect is involved .
we thus complete the formalism of pdm , in which the defect concentrations are determined by solving eqs.([eq : pdm1]@xmath32[eq : pdmx ] ) . since no interaction among defects has been taken into account in pdm , the total gibbs free energy of a defective system
is a linear superposition of the contribution of each individual defect .
that is , @xmath33 where @xmath34 is the gibbs free energy of the defect - free matrix .
the defect concentration @xmath35 runs over @xmath36 $ ] , @xmath37 $ ] , @xmath38 $ ] , and @xmath39 $ ] , respectively , with the superscript `` @xmath40 '' indicates the corresponding value in a defective reference system , and @xmath41 . from eq.([eq : gtot ] ) thermodynamic quantities as a function of stoichiometry deviation @xmath29 can be derived .
in defective crystals , distinction of the physics mainly originates from the predominant defect species . thanks to the exponential dependence of defect concentrations on the formation energy , most regions in the phase space spanned by temperature , pressure , and chemical composition ( @xmath4-@xmath3-@xmath29 )
are dominated by only one type of defect .
one can then use the concept of pseudo phase to simplify the description of defective ( non - stoichiometric ) materials.@xcite in this picture , each pseudo phase corresponds to a region that is governed by a homogeneous distribution of a _ single _ type of defect . here
no effects of migration and creation or annihilation of defects are considered , which is justified if we focus mainly on the _ long - time averaged _ properties only .
with variation of the thermodynamic conditions of @xmath4 , @xmath3 , and @xmath29 , the predominant defect might change from one type into another , _
i.e. _ , pseudo phase transition ( ppt ) might take place .
physical quantities that affected by defects also change rapidly along this transition.@xcite from this perspective , the physics of defects can be greatly simplified to that of each individual pseudo - phases and their respective behavior at the ppt .
it is thus important to understand the extent of the control region of each pseudo phase , namely , the pseudo phase diagram ( ppd ) . from eqs.([eq : pdm1]@xmath32[eq : pdmx ] ) , it is evident that such a diagram is completely described by the energetics of each defect .
conversely , if we know the ppd , then constraints on defect formation energies can be established .
it is natural to define the transition point of a ppt at where defect concentration increasing / decreasing to the half of its saturate value , then the corresponding pseudo - phase boundary ( for example in uo@xmath30 ) is determined by the following equation @xmath42 for a transition between @xmath43 and @xmath44 when @xmath45 , and @xmath46 between @xmath47 and @xmath48 when @xmath49 ( only point defects are considered ) . because all @xmath50 are functions of @xmath4 and @xmath3 ( and also depend on @xmath29 via defect interactions , which we discarded here but
will discuss in detail below ) , solutions of eq.([eq : ppbxp ] ) and ( [ eq : ppbxn ] ) provide a set of constraints on formation energy of intrinsic defects .
calculated with crocombette ( jpc)@xcite and gengs@xcite formation energies of defects .
the constraints on intrinsic defect formation energies when the pseudo transition pressure bounded between 0 and 40gpa are also shown.,width=3 ] this subsection is devoted to a realistic system of uo@xmath0 , where involved parameters are obtained by density functional theory calculations . since
early experiments were driven by application of uo@xmath0 as a nuclear fuel , most investigations focused mainly on formation energy of intrinsic defects , which closely relates to the parameter @xmath16 in eq.([eq : colde ] ) . to be consistent with conventional notations , we use @xmath51 , @xmath52 , and @xmath53 to denote the formation energy of schottky defect , oxygen frenkel pair , and uranium frenkel pair at zero pressure and temperature , respectively . based on the defect energetics calculated by geng _
et al . _ for uranium dioxide,@xcite the ppd on the @xmath3-@xmath29 plane
are evaluated and shown in fig.[fig : ppd ] .
it can be seen that in the region of @xmath49 an increasing of temperature from 300 to 2500k has little influence on the pseudo boundary .
however , in the @xmath45 side , such a size of change in @xmath4 leads to an increase of the pseudo transition pressure about 5gpa . except that
, the impact of @xmath29 on ppd at the pdm level of approximation is small , and with an opposite trend for the hypo- and hyper - stoichiometry region .
there are a few other theoretical assessments of the defect energetics for uo@xmath0 available in literature.@xcite unfortunately in most cases only formation energy at 0gpa and 0k were given , from which , however , one can not determine the pseudo transition pressure because the information about the variation with temperature and pressure is lost . under an assumption that the compression behavior and phonon contribution are the same for all of these calculations , which is a simple but reasonable approximation , we can estimate the corresponding pseudo transition boundary by adjusting the @xmath16 in eq.([eq : colde ] ) accordingly to yield the respective formation energy at 0gpa and 0k using geng s equation of state.@xcite here we choose the data of j. p. crocombette ( jpc ) for the purpose of comparison , since it is a typical one that has considered possible charge states of defects , thus produced a formation energy of oxygen frenkel pair and schottky defect that seems in a good agreement with experimental estimates.@xcite by adjusting @xmath16 to reproduce the @xmath51 and @xmath52 of jpc , we obtained the estimated ppd of jpc s data .
note here we have made an assumption that the charge state of each defect is fixed during compression or heating , and the formation energy of uranium interstitial was taken from geng s data because in jpc s work no value for this defect type was given .
the ppd calculated with jpc s formation energy is drawn in fig.[fig : ppd ] for comparison . at @xmath49 side ,
the transition from @xmath48 to @xmath47 takes place at a much higher pressure .
this is reasonable since jpc s data has a lower formation energy for @xmath48 than geng s evaluation , which gives rise to a stronger stability of this defect . at the @xmath45 side ,
uranium vacancy was predicted to be the predominant defect at low pressure , and the transition to @xmath44 occurs at about 8gpa .
this result is consistent with previous pdm evaluations at zero hydrostatic pressure , where @xmath43 was predicted to be the major defect component .
nevertheless , prevailing @xmath43 is contradictory to the experimental observation that oxygen defect should dominate this region , indicating that the experimental estimate of the defect formation energy might be inconsistent in itself.@xcite this difficulty could stem from the procedure of extraction defect energetics from diffusion measurements . usually the employed physical models were very simple and might lead to inaccurate explanation of the measured data.@xcite on the other hand , as mentioned above , ppd provides valuable information about possible range of the defect formation energy . for uo@xmath0 , the ppt in fluorite structure ( if exists ) should be bounded between a pressure range of 0 and 40gpa ,
because at higher pressures uo@xmath0 transforms into @xmath54 phase,@xcite which is followed by an iso - structural transition,@xcite and finally converges to isotropic @xmath55 structure@xcite according to recent theoretical predictions . under this restriction
, the formation energy of intrinsic defects should satisfy @xmath56 when @xmath49 , and @xmath57 when @xmath45 for a ppt to occur between a pressure of 0 and 40gpa .
these constraints are marked in fig.[fig : ppd ] together with dashed lines that indicate the corresponding bounded range of pressure . .
, width=3 ] theoretical assessment explicitly suggests that @xmath43 definitely becomes unfavorable under compressional conditions .
therefore the experimental observation that @xmath44 prevails at @xmath45 region implies that there should be no ppt from @xmath43 to @xmath44 at any pressures greater than zero
. then from eq.([eq : ppd2 ] ) one gets @xmath58 which puts a strong constraint on possible value of defect formation energies . for example
the experimental estimates of @xmath52 lie in between @xmath59ev and @xmath51 between @xmath60ev.@xcite if we take @xmath51 as 7.0ev , then @xmath52 must be less than 3.5ev .
this value is , however , incompatible with the most reliable experimental assessments.@xcite on the other hand , if we take the neutron scattering measurement@xcite of 4.6ev as a reliable estimate for @xmath52 , then @xmath51 must be greater than 9.3ev .
this in turn disqualifies most theoretical estimates with charged defects.@xcite in a word , all of these indicate that we need further scrutiny on these estimates , and any alternative and/or complemental information on defect energetics are decisive to reach the final conclusion
. inequalities of eqs.([eq : ppd1]@xmath32[eq : ppd2 ] ) cover most range of the experimental estimates , as shown in fig.[fig : tp ] .
thus it can provide new understanding about this issue if we can measure the compression - driven ppt of point defects experimentally .
defective behavior of materials at high pressure is determined by the variation of defect formation enthalpy with compression .
it is also affected by possible structural transitions of the matrix .
above discussion elaborated what might happen in a compressed non - stoichiometric uo@xmath0 .
in other materials , however , much more complex phenomena can be expected . from a theoretical perspective , it is helpful to explore all possibilities in order to grasp the general feature of the physics of defects . in the simple pdm approximation , the physics
is mainly determined by the parameters appeared in eq.([eq : colde ] ) , _ i.e. _ , the value of @xmath16 , @xmath17 , and @xmath18 of each defective configurations that were employed to derive the formation gibbs free energy ( fge ) .
therefore we can arbitrarily alter these parameters to probe other interesting behaviors of defects that are allowed in theory but not in the realistic uo@xmath0 .
the first phenomenon we would like to discuss about is reentrant transition .
it is a rare type of phase change even for a conventional physical state , where one phase that has been transformed into another re - appears . on the phase diagram
the corresponding phase boundary is a reentrant curve .
analogous phenomenon can also take place in ppt , where the predominant defect species firstly changes into another type , and then transforms back .
the condition of this transition is completely governed by eqs.([eq : pdm1]@xmath32[eq : pdm3 ] ) .
put it explicitly , if the derived equation [ eq.([eq : ppbxp ] ) or ( [ eq : ppbxn ] ) ] has multiple solutions , then the corresponding ppt is reentrant . .
upper panel : energy variation ; lower panel : defect concentrations .
, width=3 ] figure [ fig : reentry ] demonstrates a virtual dioxide compound that has a reentrant ppt . in the upper panel , a graphical solution of eq.([eq : ppbxp ] ) is drawn where the solid line is the term of @xmath61 .
another term of @xmath62 is also shown as a dashed line in the figure .
the points of intersection a and b correspond to the solutions of eq.([eq : ppbxp ] ) , which also are the locations where ppt taking place .
in the lower panel , the change of defect concentrations along compression is illustrated , from which one can clearly see that cation vacancy re - appears at higher pressures .
it is necessary to point out that this result was obtained by subtracting 145gpa from the bulk modulus of all defective uo@xmath0 configurations , thus might be an artifact .
nevertheless , such a virtual model can help us acquire a profound understanding of material s behavior that having reentrant ppt , if it exists .
the resultant modifications on eos and thermodynamic properties across this transition region are interesting .
figure [ fig : pv_ppt ] illustrates the compression behavior of the same ppt as in fig.[fig : reentry ] , in which the restoring of the compression curve at high pressure end is evident .
there are two volume - collapses in the reentrant ppt at point a and b , respectively . between a and b , the curve is steeper .
this is consistent with the requirement for a pressure - driven reentrant transition to occur , namely , the intermediate phase should have a larger bulk modulus and a smaller equilibrium volume .
this condition guarantees a double volume - collapse along compression , which is a necessity for a first - order reentrant transition .
other thermodynamic properties also show discontinuous or quasi - discontinuous jump at the ppt points .
the inset in fig.[fig : pv_ppt ] draws relative variation of thermal expansivity @xmath63 with respect to its initial value along the 300k compression curve .
it deviates from the trend of the initial phase ( as the dashed line shows ) at point a and plunge to a new value , but at point b it jumps back to the previous curve , a key feature of reentrant transition .
other physical quantities , such as specific heat and compressibility , demonstrate similar characteristics . .
inset : the relative variation of thermal expansivity .
, width=3 ] the picture of pseudo - phase of defects is only valid when the dominating region of the associated defect type is well defined .
for point defects at low temperature , it is usually the case .
however , with elevated temperature and/or when complex defect clustering is involved , competition might lead to coexistence of different defect species , where the notation of pseudo phase could lose its physical importance . for realistic uo@xmath0 , pseudo - phase
can always be defined , whether oxygen clusters ( _ e.g. _ , cot - o cluster ) involved or not . at high temperatures ,
the zone of pseudo phase boundary becomes wide , and renders the ppt as a smooth crossover.@xcite in spite of this , the material behavior still can be understood within the framework of pseudo phase . in some conditions , however , a situation that one defect species appears but never gains the dominant role might be possible .
this will completely invalidate the picture of pseudo phase , and a detailed analysis of defect concentrations becomes necessary ( in contrast to this , the defect concentrations are determined by stoichiometry deviation if pseudo phase can be applied ) . , see text for details . , width=3 ] an example of this is presented in fig.[fig : coext ] , which was generated with jpc s formation energy data.@xcite in addition to that , the value of cot - o cluster@xcite was also reduced by 4.865ev , namely , we have artificially decreased the stability of oxygen clusters .
the result is that oxygen interstitial o@xmath64 gets some promotion , but not enough to become the major one , and persists only within a narrow pressure range .
it is evident from fig.[fig : coext ] that in this virtual system u@xmath64 and cot - o ( at least for most pressure range ) can be depicted using pseudo phase , and there is a ppt from u@xmath64 to cot - o at a pressure of 7gpa . on the other hand , o@xmath64 , which appears from 20 to 50gpa , depletes the concentration of cot - o slightly .
but it never gains the dominance , and the phenomenon must be taken as a defect coexistence rather than being a ppt . to understand the difference between these is crucial for a correct description of defective materials using the concept of pseudo phase .
above discussion is based on the approximation of pdm , which is valid at the dilute limit . with an increase of defect concentrations
, however , interaction between defects becomes significant .
this could lead to a severe deviation from the prediction of pdm . on the other hand ,
non - stoichiometry might affect the eos and energetics of defects , thus modify the ppd . to obtain the correct defect behavior near the stoichiometry ,
a close cooperation between experimental measurements and theoretical analysis is necessary . for that purpose ,
a thorough and comprehensive understanding of defect interaction is crucial .
this is because with such an accurate information , we can deduce the formation energy at dilute limit from the measured ppd at finite stoichiometry deviations .
it is worthy of mentioning that the same extraction procedure can be done using the simple pdm , but the resultant error is usually quite large . in this part of the paper
, we will first derive a general formalism that includes defect interactions , then we will show that the conventional pdm can be adapted to take these interactions into account effectively with a simple constant mean - field approximation , which greatly expands the applicability of pdm .
if one is interested mainly in macroscopic properties of a defective material that averaged over a long enough time scale , the most important contribution comes from defects in static distributions . for a classical crystal at finite temperature ,
defects have nonzero probability of migrating between available sites .
this alters the dynamics and transport property of the system .
nevertheless , diffusion does not modify the overall macroscopic properties very much as long as the distribution of defects is still in an equilibrium state .
this is because migration is a transient process and in most time defects are trapped in their equilibrium sites . in a migration process , defects induce dynamic deformations in the local lattice , and scatter with phonon .
this effect might drive phonon away from its equilibrium distribution .
however , considering that this change is very small compared to thermal fluctuations , and our interested time scale is many orders longer than the relaxation time of phonon , it is justified to ignore this effect .
the main assumptions in the theory include : ( a ) we work on a lattice model , ( b ) only static distribution of defects is considered , ( c ) the distribution must be homogeneous , ( d ) dynamic effects due to defect migration are ignored , ( e ) no defect creation and annihilation are considered , ( f ) when defect clusters are involved , taking them as single objects , namely , ignore association and/or dissociation effects . with these assumptions ,
let us consider a grand canonical ensemble on a lattice in which the number of particles fluctuating around the average value .
assume there are in total @xmath65 sublattices , which are occupied by @xmath15 species of particles .
using @xmath66 denotes the number of @xmath67-th kind of particle that siting on the @xmath1-th sublattice , where @xmath1 can take any value from 1 to @xmath65 and @xmath67 varies from 1 to @xmath15 .
it is evident that each system in the ensemble is characterized by the occupation of particles on the lattice , and thus the ensemble can be grouped according to @xmath68 .
that is to say , each system can be labeled uniquely by a set of @xmath68 together with an auxiliary index @xmath69 , where @xmath69 runs over all possible configurations that has the same @xmath68 . in this way , if using @xmath70 denotes the probability of configuration @xmath71 in the ensemble , where @xmath71 has been shortened as @xmath72 for brevity , then the gibbs free energy of the ensemble is @xmath73 .
\label{eq : gibbs1}\ ] ] here @xmath74 is the gibbs free energy of an individual configuration , and the contribution of configurational entropy has been separated out and presents as the second term in eq.([eq : gibbs1 ] ) .
the thermodynamically equilibrium state is achieved when @xmath75 takes a minimum .
the normalization condition requires @xmath76 . if @xmath77 denotes the total number of the @xmath67-th particle in the ensemble , then @xmath78 .
introducing lagrange multipliers @xmath79 and @xmath80 , the minimization equation of @xmath75 becomes @xmath81 making use of eq.([eq : gibbs1 ] ) , this leads to @xmath82 where the partition function @xmath83 .
it can be shown that @xmath79 is just the chemical potential of the @xmath67-th type of particle .
the free energy of the @xmath72-th configuration , @xmath84 , which consists of the cold crystal energy at 0k and phonon contributions to the internal energy and entropy ( as well as a term of @xmath85 , if at finite pressure ) , can be calculated with modern first - principles methods . since defects in an individual configuration manifest as
an imperfect occupation of the lattice sites , the equilibrium thermodynamics of defects is therefore completely described by eq.([eq : rho ] ) , in which the contribution of defect interactions arises from @xmath84 .
when the energy scale of temperature is much smaller than the formation energy of defects , which usually is the case for most applications , the partition function is dominated by the perfect occupation ( _ i.e. _ , the ground - state ) , namely @xmath86 , where the subscript 0 denotes the groundstate . in this situation
eq.([eq : rho ] ) reduces to @xmath87 where the fge of configuration @xmath72 is given by @xmath88 .
for single point defect @xmath89 .
thus the fge of intrinsic defects such as frenkel pairs and schottky defect can be derived straightforwardly from eq.([eq : rhoq ] ) , and have the same form as eqs.([eq : gfp]@xmath32[eq : gs ] ) .
furthermore , equations similar to pdm [ eqs.([eq : pdm1]@xmath32[eq : pdm3 ] ) ] also can be constructed easily using eq.([eq : rhoq ] ) , indicating that the simple formalism of pdm is much more flexible than what it was originally proposed .
it is necessary to point out that for a system defined on a lattice , there is an orthogonal and complete basis set called correlation functions , which are an alternative but powerful representation of all possible occupation configurations of particles on the lattice using increasingly complex point sets that ranging from single point to nearest pairs and bigger clusters.@xcite any functions defined on the lattice can be expanded with such a basis . for defective crystals ,
the matrix and interstitial sites define such a lattice naturally .
if we further introduce a special species of _ white atom _ , and regard all vacancies as that occupied by white atoms , then such orthogonal and complete correlation functions @xmath90 can be defined . in this way
, the probability for a configuration to appear becomes @xmath91 and the gibbs function of eq.([eq : gibbs1 ] ) can be rewritten as @xmath92 with the configurational entropy given by @xmath93 $ ] , and the interaction strength of cluster @xmath94 .
it is obvious that eq.([eq : g_alloy ] ) has the same form as the theory for alloys.@xcite namely , both alloying and defects on a lattice can be described by the same unified theoretical framework . within this method ,
the interaction strength of clusters can be evaluated with cluster expansion method using _ ab initio _ total energy calculations,@xcite and the configurational entropy may be evaluated by cluster variation method.@xcite for a point defect on an infinitely large lattice ( the dilute limit ) , the formalism discussed above naturally leads to the conventional pdm.@xcite when defect concentration is finite but interaction between them is weak and can be ignored safely , the same conclusion holds .
this is because for a configuration containing @xmath95 non - interacting point defects , the corresponding fge is just @xmath95 times of that of a single one , and then eq.([eq : rhoq ] ) gives rise to @xmath96 , which is exactly the result of pdm . alternatively , for non - interacting defects , the presence of a defect has no influence on others , we therefore can isolate a defect by cutting it and the associated local lattice out from the matrix , and then extend the surrounding lattice to an infinite range .
this operation keeps the defective behavior .
it maps non - interacting defects with finite concentrations onto a group of systems at the dilute limit , which justifies the application of pdm . when interactions between defects are substantial , it is almost impossible to isolate a defect from others . at a condition that the distribution of defects is homogeneous , however , we can approximate the interaction by a constant mean - field . in this approximation ,
the defects and the associated local lattice environment that are cut out from the matrix keep the original size , and subject to a field that takes a role of modeling the interactions with other homogeneously distributed defects that have been removed . to construct accurately an environmental field of such kind is difficult , if not impossible . for practical purpose
, we may simplify it by using regularly distributed defects to simulate the field approximately .
this sacrifices the rigidity of the theory , but makes the problem more tractable .
what one needs to do now is to periodically repeat the piece of defective lattice that has been cut out from the matrix along the three dimensional lattice vectors . in this way
the periodical images of the defect take the role of modeling the homogeneous environment produced by other defects .
just one such configuration of course can not capture the whole features of the defect interactions . by averaging over all possible regular defective distributions ,
nevertheless , one eventually can reach a converged result .
generally the free energy of a system in the ensemble can be written as @xmath97 where @xmath35 is the concentration of defect @xmath1 in this system .
if only up to linear terms are kept , then eqs.([eq : gibbs1 ] ) and ( [ eq : rhoq ] ) reduce back to the conventional pdm .
the terms of higher order describe effective defect interactions , and should be important for any real materials with high defect density . on the other hand ,
if we know the values of parameter @xmath98 , @xmath99 , @xmath7 , and @xmath100 , the free energy ( as a function of defect concentration @xmath35 ) of any defective configuration can be evaluated from eq.([eq : gibbs2 ] ) directly . to determine these parameters , one can solve eq.([eq : gibbs2 ] ) by a least square fit method using _ ab initio _ calculated @xmath84 of a set of configurations .
furthermore , since the @xmath50 in eq.([eq : rhoq ] ) can be rewritten as @xmath101 where the first term at the right hand side represents the contribution of non - interacting defects , and the second term arises from defect interactions , eq.([eq : rhoq ] ) then leads to @xmath102 where @xmath103 is the probability predicted by conventional pdm which gives a defect concentration of @xmath104 ( determined by @xmath105 ) . for example
, if there are @xmath106 defects of @xmath1-th type appearing in the configuration @xmath72 , where @xmath107 and @xmath108 is the total available sites for that defect , pdm gives @xmath109 . because of the structure of @xmath110 as shown in eqs.([eq : gibbs2 ] ) and ( [ eq : dg ] ) , @xmath70 can be factorized into the same form as @xmath111 .
namely , @xmath112 we finally get @xmath113 this explicitly demonstrates that the simple formalism of pdm is still valid even when defect interactions present , as long as the defect distribution is homogeneous . the effect of interaction is to modify the defect concentrations in a constant mean - field way ( here _ constant _ means that the interaction has been averaged over the whole configurational space so that no dependence on the distance between defects presents explicitly ) , and the formation energy of a single point defect @xmath114 has to be changed from its dilute limit value @xmath115 to @xmath116 and the defect concentration equations also become @xmath117 .[tab : cohee ] first - principles results for the energy curve of uo@xmath118 , where @xmath29 is the deviation from the stoichiometric composition of uranium dioxide , @xmath24 is the total number of atoms in the simulation cell .
@xmath16 ( in ev ) , @xmath119 ( in @xmath120 ) , and @xmath17 ( in gpa ) are the cohesive energy per atom , the equilibrium lattice parameter of the effective cubic cell , and the zero pressure bulk modulus , respectively . [
cols="<,^,^,^,^,^,^,<",options="header " , ] according to ref . and above discussions , pseudo phases in uo@xmath0 are well defined : at most interested thermodynamic conditions only one type of defect presents , and all other components are suppressed completely .
this implies that only the diagonal terms in eq.([eq : gibbs2 ] ) make sense .
namely , only interactions between the same kind of defect need to be considered , which greatly reduces the number of _ ab initio _ calculations that are required for extraction of the interaction strengths @xmath7 and @xmath100 . using configurations with different simulation cell size of @xmath121 , @xmath122 , and @xmath123 of the cubic fluorite unit
, we extracted the defect interaction strength by solving a set of equations of eq.([eq : gibbs2 ] ) .
the employed energy curves for the smallest cell are listed in table [ tab : cohee ] ( in which the defective structures are labeled follwoing the same rule of ref . ) , and others are taken from table i in ref .. in particular , the results in table [ tab : cohee ] were calculated using vasp code , with the same lsda+__u _ _ setting as in ref .. all structures were fully relaxed at a series of fixed volumes . since the supercell size of these structures is relatively small , 36 irreducible @xmath124 points were employed to ensure the total energy convergency . with the effective pdm of eq.([eq : gpdm ] ) , it is not necessary to work on the ensemble average of eq.([eq : gibbs1 ] ) any longer .
instead , the problem changes to `` how the effectively independent defects distribute on the lattice '' . for a purpose of investigating the compression behavior of defects ,
it is helpful to employ a reference supercell , and normalize all involved energetics with respect to it . in doing so
, however , the number of defects might no longer be an integer .
a re - scaling procedure is thus required when evaluating the formation energy of a _ single _ defect .
let @xmath125 be the defect contribution in eq.([eq : gibbs2 ] ) that is evaluated in the reference cell , _
i.e. _ @xmath126 , with a defect concentration equals @xmath127 . the number of unit cells in a supercell which contains _ one _ and _ only one _ of this type of defect is @xmath128 , where @xmath129 is the number of available sites for this defect in a unit cell .
then the energy difference for creating a defect is @xmath130 , where @xmath131 is the number of unit cells making up the reference supercell . in this way , the formation energy of a frenkel pair for @xmath132 species becomes @xmath133 and the schottky defect formation energy is @xmath134 here @xmath24 is the total number of atoms in a defect - free reference cell . for a @xmath121 supercell of fluorite uo@xmath0 , @xmath135 and @xmath136 . also for a cubic fluorite unit
, @xmath129 takes 8 for o@xmath137 , and 4 for o@xmath64 , u@xmath137 , and u@xmath64 , respectively .
these formulations , together with eq.([eq : df_int ] ) , allows for extrapolating the intrinsic defect formation energy to the dilute limit by decrease the defect concentration to an arbitrarily small value .
the obtained results are summarized in table [ tab : ef ] .
we can see that a @xmath121 cell is not big enough to converge the formation energy to the dilute limit .
the deviation from the extrapolated value is less than 1ev .
the largest one is oxygen frenkel pair in which @xmath138 reaches a value of 0.6ev .
it is at the same level of the finite - size correction of charged defects,@xcite where a value of about 0.6ev was also obtained for o_fp .
this good agreement demonstrates that our treatment on defect interaction is at least qualitatively correct . ) and compressibility ( @xmath139 ) with stoichiometry deviation @xmath29 at different level of approximations .
notice the sharp tips at @xmath140 due to defect interactions .
, width=3 ] inclusion of defect interactions into the pdm makes it possible to study the fine behavior over the whole stoichiometry .
figure [ fig : alpha ] shows the relative variation of thermal expansivity @xmath63 and compressibility @xmath139 as a function of stoichiometry deviation @xmath29 at 300k and 0gpa , in which the solid points denote the exact value of the configurations that were employed to extract the interaction strength .
it is evident that the linear approximation of eq.([eq : gibbs2 ] ) ( namely the conventional pdm ) fails to reveal the fine behavior of non - stoichiometric uo@xmath0 .
far from the point that was used to approximate the defect formation energy , it deviates from the exact value drastically . on the other hand , both quadratic and cubic approximations predict a curved variation of physical quantities correctly .
it is interesting to note that the sharp tip appearing at the stoichiometry is very similar to the `` w '' shape anomaly in alloys.@xcite nevertheless , the underlying physics is different .
here it mainly originates from two facts : ( a ) the curvature due to defect interactions , and ( b ) the predominant defect type at the hyper- and hypo - stoichiometry sides are different , which gives rise to different variation trends of the physical quantities .
from this perspective , measuring the departure of relevant physical quantities from a linear behavior near the stoichiometry would reveal the strength of defect interactions , and constrain the application range of the conventional pdm . above discussions revealed the power of including defect interaction into statistical mechanics model such as pdm .
though by comparison with available experimental data one can assess the validity of our approach , due to the uncertainty of these data as mentioned in sec.[sec : intr ] and [ sec : pdm ] , a quantitative validation has to be made with direct _ ab initio _ calculations .
to this end , we made an independent calculation on a defective configuration of @xmath141 , using a supercell of @xmath142 of the cubic fluorite unit , in which one oxygen atom has been removed to create one oxygen vacancy .
the obtained energetics are listed in table [ tab : cohee ] , and the calculated thermal expansivity and compressibility are compared with that of effective pdm in fig.[fig : alpha ] as the open square points .
note that in this separate calculation we employed only the experimental observation that oxygen vacancy prevails in uo@xmath143 , and did not invoke any other approximations .
therefore the good agreement between these two results as shown in fig.[fig : alpha ] provides a solid verification of the validity of our constant mean - field treatment of defect interactions for homogeneously distributed defects , as well as the effective pdm that based on it .
uo@xmath0 is a semiconductor with a finite energy gap , it is possible that defects in it are charged rather than being neutral . except very few studies,@xcite most _ ab initio _
investigations on defective uo@xmath0 assumed a neutral simulation cell , as we did here . in principle
, such `` neutral '' calculations do not correspond to literally neutral defects , since local transfer of electrons might lead to a partial charge of the defects .
nevertheless , finite size of the simulation cell imposes a constraint on the charge redistribution , and thus defect might not reach its full charge state .
this problem becomes very severe when at the dilute limit or near the stoichiometry , where defects can be fully charged only by exchanging electrons with valence / conduction bands , which is a kind of global charge redistribution . for large stoichiometry deviation , however , defects interact with the matrix strongly , leading to deep defect levels which are either at near the gap middle or hybridization with valence / conduction bands . in both cases a neutral defect could be expected because it is difficult to _ ionize _ the defect at low temperatures in the first case and no charged defect can be supported in the second one . for oxygen clusters ,
crocombette argued that charged state makes them unfavorable at the stoichiometry.@xcite it is reasonable .
actually , `` neutral '' calculations also indicated that oxygen clusters have negligible concentrations when near the stoichiometry.@xcite putting these information together , we ascertain that it should have no defect clustering when @xmath144 . but this does not mean that oxygen clustering is negligible at non - stoichiometry .
the experimental evidence for such clustering was in fact observed at large values of @xmath29,@xcite which is compatible with recent neutral calculations that predicted prevailing cot - o clusters at hyper - stoichiometry region . on the other hand , _ ab initio _ electronic structure revealed that the defect levels of cot - o cluster hybridize strongly with the valence band of the matrix,@xcite which implies that the cluster might be `` neutral '' , or at least these `` neutral '' results should partially reflect some physical reality . in these considerations ,
crocombette s conclusion about charged oxygen defects@xcite might lose the relevance when far from the stoichiometry .
but overall the charged state of defects in uo@xmath0 is still an open issue . in the derivation of the general formalism of defects in sec.[sec : defect interactions ] , we did not consider the charged state . to include
this is straightforward .
one just needs to add an additional index to each type of defect to mark its charge state , and include the chemical potential of free electrons to take the charge contribution at the fermi level into account .
eq.([eq : gibbs1 ] ) then becomes @xmath145 , \label{eq : g_charge}\ ] ] where @xmath146 is the total charge of the system , and @xmath147 the fermi level . from this expression , the effective pdm which includes both defect interactions and variable charge state of defects
can be derived easily . .
, width=3 ] although ppt and the corresponding boundaries can be employed to constrain / extract the dilute limit of defect formation energy
an important quantity for understanding the stoichiometric behavior , to measure these boundaries is not a trivial work . at big stoichiometric deviation ,
the volume change at ppt is prominent , thus it can be detected by measuring the quasi - discontinuous jumps in the eos of defects.@xcite figure [ fig : pv ] plots a compression curve of uo@xmath148 along a hugoniot shocked from 300k and 0gpa .
the volume collapse due to ppt from u@xmath137 to o@xmath64 ( here we ignored oxygen clustering ) is evident and therefore detectable . on the other hand , this pseudo transition pressure depends sensitively on the formation energy of @xmath44 : it spans over a wide range of 40gpa when there is a change in @xmath149 about 4.0ev .
this property guarantees a good precision for the constraints on defect energetics . at small value of @xmath29
, however , the volume jump would be too weak to be perceptible .
this is usually the case when @xmath150.@xcite in these cases , we can not locate the ppt via measuring thermodynamic or mechanical quantities .
however , since ppt changes the predominant defect species and thus the position of the defect level within the energy gap , transport properties are also modified .
we therefore can detect the occurring of a ppt by measuring the sudden changes in electrical conductivity ( or optical properties).@xcite this method has high sensitivity so that allows us to access the vicinity of the stoichiometry . ) , isothermal bulk sound velocity , and specific heat at constant pressure ( @xmath151 ) along a hugoniot that shocked from 500k and 0gpa in non - stoichiometric uo@xmath0 when defect concentrations have an instantaneous response to thermal fluctuations .
the solid circle ( at the bottom left corner ) marks the experimental bulk sound velocity measured at ambient conditions in perfect uo@xmath0 .
, width=3 ] in above discussions and also in ref .
, we froze the defect concentrations when evaluating the thermodynamic quantities .
it is a theoretical requirement for the first derivatives of the gibbs function , such as volume and entropy .
but for higher order derivatives of the gibbs function , it has no reason to do so because they are also defined by thermodynamic relations . in practice , however , a justification for this operation can be made .
this is because for a classical crystal , the change of defect species can proceed via only atomic diffusion , which is a very slow process , and thus no defect can respond to rapid thermal fluctuations .
then an interesting question arises , that is what about if defects can instantaneously respond to any disturbances ?
simple analysis shows it might be fantastic . at first the magnitude of anomalies due to ppt
would be amplified greatly , thus ease the difficulty in ppt detections .
figure [ fig : cpp ] demonstrates this effect on the thermal expansivity @xmath63 , bulk sound velocity , and specific heat at constant pressure @xmath151 .
the influence can be fully comprehended by comparing with the fig.2 in ref . , where defect concentrations were fixed when evaluating these quantities .
secondly , the compressibility would diverge at zero temperature .
considering the relationship between sound velocity and the compressibility , this implies a vanishing sound velocity ( and the bulk modulus ) in the vicinity of a ppt at low temperatures if defects have instantaneous response . on the phonon spectrum , it would manifest as an abnormal softening in acoustic branches at long - wave length ( _ i.e. _ , @xmath152 point in the reciprocal space ) .
this observation is tantalizing . but
can it be true ?
we can not answer it yet .
classical atomic diffusion of course can not lead to a rapid response .
but what if for a quantum crystal ? in a quantum world particles are described by a wave - function
. if the wavefunctions of all defects are in a coherent state , then an instantaneous and simultaneous change of the defect species might be possible.@xcite in this mechanism the requirement for a large scale atomic diffusion has been removed , and a single change in defecton state is enough for a quantum pseudo transition to occur .
nevertheless , much more theoretical works are necessary in order to pin down this possibility definitely .
a general formalism for the thermodynamics of defects in a crystal was derived based on the statistics of a grand canonical ensemble on a lattice . by introducing idle white atoms for vacancy and extending the sublattice of interstitial sites ,
this formalism has the same form as the lattice theory for alloys and compounds a reflection of the unified physics underlying these seemingly different systems . with an approximation of constant mean - field
, this theory reduces to an effective point defect model in which defect interactions are included by an auxiliary field , whereas each individual defects are treated as independent . in this way
, we mapped a _ many - body _ defect system onto a _
single _ defect system by coupling it with an effective external field .
if ignoring this field , the conventional pdm is recovered .
this generalization greatly expands the applicability of the simple pdm . in order to explore the full content in this theory
, we also studied possible reentrant ppt and multi - defect coexistence with virtual systems . using pdm
, we investigated the possibility of constraining defect energetics by measuring pseudo phase boundaries . by calculating the possible ppt between interested defects
, we showed that the experimental estimates available in literature , as well as variety theoretical assessments , on defective energetics of uo@xmath0 , are not fully consistent . on the other hand ,
the range of energetics constrained by the ppts overlaps with these estimates largely , therefore has a potential to reduce the inconsistency in these data . by including defect interactions into pdm
, we demonstrated that the information obtained at finite stoichiometry deviation can be extrapolated to the dilute limit .
finally , we investigated the fine behavior of thermal expansivity and compressibility in the vicinity of the stoichiometry of defective uo@xmath0 , and some relevant issues of charged defects , detection of ppt , and possible instantaneous response of defectons in a quantum crystal are briefly discussed , in which we highlighted the detection of ppt by measuring electrical conductivity when near the stoichiometry , and the complexity arising from possible charged state of defects . through these investigations
, we clearly demonstrated that it is valuable to explore the whole non - stoichiometric range in order to acquire a comprehensive understanding about a defective material thoroughly .
99 f. agullo - lopez , c. r. a. catlow , and p. d. townsend , _ point defects in materials _
( academic press , london , 1988 ) . o. t. srensen , _ non - stoichiometric oxides _ ( academic , new york , 1981 )
matzke , j. chem .
faraday trans . 2 * 83 * , 1121 ( 1987 ) .
g. e. murch and c. r. a. catlow , j. chem .
faraday trans . 2 * 83 * , 1157 ( 1987 ) .
t. g. stratton and h. l. tuller , j. chem .
faraday trans . 2 * 83 * , 1143 ( 1987 ) .
r. tetot and p. gerdanian , j. phys .
solids * 46 * , 1131 ( 1985 ) .
k. clausen , w. hayes , j. e. macdonald , r. osborn , and m. t. hutchings , phys .
lett . * 52 * , 1238 ( 1984 ) .
j. p. crocombette , f. jollet , t. n. le , and t. petit , phys .
b * 64 * , 104107 ( 2001 ) . l. petit , a. svane , z. szotek , w. m. temmerman , and g. m. stocks , phys . rev .
b * 81 * , 045108 ( 2010 ) .
s. l. dudarev , g. a. botton , s. y. savrasov , z. szotek , w. m. temmerman , and a. p. sutton , phys .
status solidi a * 166 * , 429 ( 1998 ) .
i. d. prodan , g. e. scuseria , and r. l. martin , phys .
b * 73 * , 045104 ( 2006 ) ; * 76 * , 033101 ( 2007 ) . h. x. song , h. y. geng , and q. wu , phys . rev .
064110 ( 2012 ) . h. y. geng , y. chen , y. kaneta , and m. kinoshita , phys .
b * 75 * , 054111 ( 2007 ) . h. y. geng , h. x. song , k. jin , s. k. xiang , and q. wu , phys . rev .
b * 84 * , 174115 ( 2011 ) .
y. geng , h. x. song , and q. wu , phys .
b * 85 * , 144111 ( 2012 ) .
f. gupta , a. pasturel , and g. brillant , phys .
b * 81 * , 014110 ( 2010 )
. d. a. andersson , t. watanabe , c. deo , and b. p. uberuaga , phys .
b * 80 * , 060101(r ) ( 2009 ) . d. a. andersson , f. j. espinosa - faller , b. p. uberuaga , and s. d. conradson , j. chem . phys . * 136 * , 234702 ( 2012 ) .
d. a. andersson , j. lezama , b. p. uberuaga , c. deo , and s. d. conradson , phys .
b * 79 * , 024110 ( 2009 ) .
h. y. geng , y. chen , y. kaneta , m. iwasawa , t. ohnuma , and m. kinoshita , phys . rev .
b * 77 * , 104120 ( 2008 ) . h. y. geng , y. chen , y. kaneta , and m. kinoshita , phys .
b * 77 * , 180101(r ) ( 2008 ) . h. y. geng , y. chen , y. kaneta , and m. kinoshita , appl . phys .
lett . * 93 * , 201903 ( 2008 ) .
m. freyss , t. petit , and j. p. crocombette , j. nucl .
* 347 * , 44 ( 2005 ) .
m. iwasawa , y. chen , y. kaneta , t. ohnuma , h. y. geng , and m. kinoshita , mater .
47 * , 2651 ( 2006 ) .
b. dorado , g. jomard , m. freyss , and m. bertolus , phys .
b * 82 * , 035114 ( 2010 ) . j. yu , r. devanathan , and w. j. weber , j. phys . : condens .
matter * 21 * , 435401 ( 2009 ) .
f. gupta , g. brillant , and a. pasturel , philos . mag . * 87 * , 2561 ( 2007 ) .
g. brillant and a. pasturel , phys . rev .
b * 77 * , 184110 ( 2008 ) .
d. a. andersson , b. p. uberuaga , p. v. nerikar , c. unal , and c. r. stanek , phys .
b * 84 * , 054105 ( 2011 ) .
p. nerikar , t. watanabe , j. s. tulenko , s. r. phillpot , and s. b. sinnott , j. nucl . mater . * 384 * , 61 ( 2009 ) .
j. p. crocombette , d. torumba , and a. chartier , phys .
b * 83 * , 184107 ( 2011 ) .
j. p. crocombette , phys .
b * 85 * , 144101 ( 2012 ) .
p. larson , w. r. l. lambrecht , a. chantis , and m. van schilfgaarde , phys .
b * 75 * , 045114 ( 2007 ) .
g. jomard , b. amadon , f. bottin , and m. torrent , phys .
b * 78 * , 075125 ( 2008 ) .
b. dorado , b. amadon , m. freyss , and m. bertolus , phys .
b * 79 * , 235125 ( 2009 ) . h. y. geng , y. chen , y. kaneta , m. kinoshita , and q. wu , phys .
b * 82 * , 094106 ( 2010 ) .
v. i. anisimov , j. zaanen , and o. k. andersen , phys .
b * 44 * , 943 ( 1991 ) .
a. i. liechtenstein , v. i. anisimov , and j. zaanen , phys .
b * 52 * , r5467 ( 1995 ) .
s. l. dudarev , g. a. botton , s. y. savrasov , c. j. humphreys , and a. p. sutton , phys .
b * 57 * , 1505 ( 1998 ) .
f. zhou and v. ozolins , phys .
b * 80 * , 125127 ( 2009 ) .
f. zhou and v. ozolins , phys .
b * 83 * , 085106 ( 2011 ) .
h. y. geng , m. h. f. sluiter , and n. x. chen , phys .
b * 72 * , 014204 ( 2005 ) . v.
l. moruzzi , j. f. janak , and k. schwarz , phys .
b * 37 * , 790 ( 1988 ) .
h. serizawa , y. arai , m. takano , and y. suzuki , j. alloys compd . * 282 * , 17 ( 1999 ) .
h. serizawa , y. arai , and y. suzuki , j. nucl . mater .
* 280 * , 99 ( 2000 ) .
h. y. geng , n. x. chen , and m. h. f. sluiter , phys .
b * 71 * , 012105 ( 2005 ) .
m. sluiter , d. de fontaine , x. q. guo , r. podloucky , and a. j. freeman , phys .
b * 42 * , 10460 ( 1990 ) .
m. h. f. sluiter , y. watanabe , d. de fontaine , and y. kawazoe , phys . rev .
b * 53 * , 6137 ( 1996 ) .
j. w. d. connolly and a. r. williams , phys .
b * 27 * , 5169 ( 1983 ) .
r. kikuchi , phys . rev . * 81 * , 988 ( 1951 ) .
b. t. m. willis , proc .
. soc . * 1 * , 9 ( 1964 ) .
b. t. m. willis , acta crystallogr . ,
a : cryst .
crystallogr . *
a34 * , 88 ( 1978 ) .
a. d. murray and b. t. m. willis , j. solid state chem . * 84 * , 52 ( 1990 ) .
|
making use of the energetics and equations of state of defective uranium dioxide that calculated with first - principles method , we demonstrate a possibility of constraining the formation energy of point defects by measuring the transition pressures of the corresponding pseudo - phase of defects . the mechanically stable range of fluorite structure of uo@xmath0 , which dictates the maximum possible pressure of relevant pseudo - phase transitions , gives rise to defect formation energies that span a wide band and overlap with the existing experimental estimates .
we reveal that the knowledge about pseudo - phase boundaries can not only provide important information of energetics that is helpful for reducing the scattering in current estimates , but also be valuable for guiding theoretical assessments , even to validate or disprove a theory . in order to take defect interactions into account and to extrapolate the physical quantities at finite stoichiometry deviations to that near the stoichiometry
, we develop a general formalism to describe the thermodynamics of a defective system .
we also show that it is possible to include interactions among defects in a simple expression of point defect model ( pdm ) by introducing an auxiliary constant mean - field .
this generalization of the simple pdm leads to great versatility that allows one to study nonlinear effects of stoichiometry deviation on materials behavior .
it is a powerful tool to extract the defect energetics from finite defect concentrations to the dilute limit . besides these , the full content of the theoretical formalism and some relevant and interesting issues , including reentrant pseudo - transition , multi - defect coexistence , charged defects , and possible consequence of instantaneous defective response in a quantum crystal , are explored and discussed .
|
physics developed from attempts to find order in the changes of spatial configurations of natural objects , such as celestial bodies . on the base of the classical mechanics , applicable in everyday situations , a picture emerged within which the relations between objects in space and time are a closed , self - explaining entity with future uniquely determined by the current spatial configuration of objects and the instantaneous rate of its change .
the subsequent discovery of chaos showed , however , that practically in every real situation the freedom of possible motions of the system , left after imposing the laws of classical mechanics , is extremely vast .
the relations between objects in space and time that are realized in reality , remain largely inexplicable . in other words , in complex situations , the constraints constituted by the laws of mechanics are extremely loose and they bring little information , making us to search for additional connections .
we are brought back to the original question - what is the order behind the changes of spatial configurations of objects around us ?
can it be that the laws of mechanics are emergent from some more fundamental processes the understanding of which would shed new light on chaos ?
is it possible that the spatio - temporal relations are but secondary and emerge on the basis of some deeper phenomena the essence of which has nothing to do with space and time ?
here it is important to notice the principal importance of the approximate character of the classical mechanics . in a complex situation approximations
may play a decisive role because classical chaos leads to sensitivity to small corrections which impact accumulates .
in particular , if these corrections have a certain `` informational direction '' , then the resultant global picture may differ qualitatively from the one expected on the basis of mechanics .
( the limitations of the classical mechanics perspective on chaos are discussed qualitatively in sec .
[ chapfive ] and quantitatively in secs .
[ chaptwelve]-[chapthirteen ] . )
it is natural to consider the corrections introduced by the quantum mechanics , but these , at least at the moment , have not led to a significantly new understanding of the questions posed above .
the fact that the information on the physical world provided to us by the classical or quantum mechanics does not say essentially anything about the order in chaos leads one to think that in the conventional approach we miss something fundamentally important , something that may play the critical role in the structure of chaos and thus of our world .
we are lacking general principles of orientation in complex situations and that s why the order that often arises in those situations looks to us accidental .
the criticality of the posed question is manifested by the almost complete absence of the general guiding principles allowing to get oriented in situations far from equilibrium . and
at the same time , the studies of complex phenomena , and the very existence of different areas of knowledge with different principles of operating with objects , create an impression that we are surrounded by the unknown principles of organization of chaos , the approach to the systematization of which is absent today .
our purpose in this article is to suggest such an approach .
the following paper is organized as follows .
its first part consisting of sects .
[ chaptwo]-[chapfive ] is devoted to the principles providing background for the theory exposed in the second and third parts
[ chapsix]-[chapten ] describe the general structure of the heds - web approach which basic notions are introduced .
. [ chapeleven]-[chapfourteen ] discuss the question of emergence of spatio - temporal relations from the purely informational `` heds - web '' . in conclusion we summarize the results and discuss some questions that the heds - web hypothesis could help resolving .
both classical and quantum mechanics are built on the basis of the postulate that space and time are fundamental , primary elements of reality . thus building our theories on the basis of space and time we from the very beginning exclude some possibilities in nature that may be precisely the ones determining the order of chaos in reality .
the postulate presumes that all that exists is subordinated to the spatio - temporal order .
in particular , all possible interactions of objects should be subordinated to this order and admit an adequate description in spatio - temporal terms . by this , from the very beginning , quite a strong assumption is made on the types of possible interactions in nature . a critical examination of this assumption demands a study of the consequences of the hypothesis that nature has a wider structure .
( it should be stressed that relative self - consistency of some approach to reality does not say yet anything about the reality itself , besides that this approach does not contradict it . in particular , it is not excluded that reality may allow several different , relatively self - consistent approaches at the same time . ) a consistent study of the question if space and time really constitute the all - encompassing , universal order , reigning unconditionally , at least , in the usual everyday situations , has received relatively little attention until now . yet , as such , the classical newtonian outlook at space - time as a rigid , unshakable structure of reality has undergone significant changes .
the general theory of relativity led to the view of space - time as a dynamical and flexible structure which geometry is not fixed a priori .
the theory , however , continues to accept space and time from the very beginning as the exhaustive order fit by everything happening in nature . in turn
, some applications of the quantum mechanics to the study of space showed that the latter may be not the basic element of the theory , but , rather , an entity emerging on the basis of interactions of primary discrete objects .
however , here as well , the considered fundamental structures should from the very beginning obey the demand to reproduce space - time as the universal order of nature , at least , in the situations experimentally achievable today .
it should be remembered that space and time are just our ways to introduce logic in the natural phenomena .
the logic of the nature itself may be more complex .
if there are phenomena subordinated to more complex types of order than the spatio - temporal one , then the logic based on the latter , is incomplete . today
, natural phenomena are considered starting from the assumption that the spatio - temporal order exhausts all possibilities of nature . in particular , the order taking place in complex situations is considered as emerging largely accidentally from the interaction of a large number of elements in space and time .
thus , from the very beginning by limiting our viewpoint by space and time we close for ourselves the possibility to consider those properties of nature that are not describable in spatio - temporal terms .
just such properties may be possessed by complex phenomena in chaos .
thus a necessity arises to have a different , wider approach allowing to take into account the possible presence in nature of various types of order .
thus the search for order in nature suggests the reconsideration of the postulate that space and time are primary notions . in the approach to this reconsideration that is suggested below
, one speaks of order and connections not contradicting the spatio - temporal ones , but extending them .
the resonant web ( `` heds - web '' ) approach , which is the subject of the second part of the paper , allows to build a physical theory not introducing the demand that all interactions of objects should from the very beginning be subordinated to the spatio - temporal order . within the frame of this approach space and time
are not the primary elements of the theory . the choice of the primary elements appeals to the fundamental , basic principle - the nature consists of interacting objects .
any real situation or experiment include complex objects ( remind that waves may also be considered as consisting of objects ) and their interactions .
these are the latter which are the primary elements of the heds - web principle .
the consideration of objects and their interactions as primary elements of the theory is essentially the return to the approach to reality preceding the creation of any system of thinking about it , with the purpose of creating a wider system . from the very beginning , in our approach to reality ,
the objects exist as such .
the objects interact and these interactions happen along certain `` tones '' , in accordance with the inner characteristics of these objects .
in addition , an object is from the very beginning limited by the types of interactions in which it is able to participate .
it makes sense to consider a certain order - a system of measurements establishing relations of a certain type between objects and convenient for describing the given `` section '' of reality .
the order on which physics is based is the spatio - temporal order ( below called also `` classical '' ) .
such an order is largely anthropocentric , being strongly connected with the task of the local survival of human in space and time .
it is good for describing some interactions and it may turn out to be incomplete for describing others .
the system of measurements connected with the classical order are measurements of distances between objects with the help of rods , that prescribe objects spatial configuration , and the correspondence between changes of those configurations and readings of physical clocks . in the light of anthropocentric nature of the classical order
, it can be expected that in nature there are interactions that are not `` caught '' with the help of measurements based on rods and clocks and their different extensions .
we suppose that these interactions lead to the randomness that we observe in chaos from the viewpoint of the `` spatio - temporal section '' ( classical chaos is deterministic and `` randomness '' here means effective randomness of a single realization of dynamics ) .
including these interactions within the structure of physics ( and thus necessarily going out of the frame of the spatio - temporal section of reality ) , we have a chance to consider randomness as a law .
thus , on the basis of what was said above , experiments based on the spatio - temporal order could miss a kind of interactions that we call `` informational resonances '' or heds ( from the hebrew `` he d '' - response , echo ) .
he d is an interaction of objects that does not belong to the spatio - temporal order , that is an interaction that can not be measured with the help of rods , clocks or their extensions .
informational resonance is in principle not describable in the spatio - temporal coordinates , constituting a purely informational interaction of objects .
belonging to a different order , he d should not decay with the spatial , and thus according to the lorentz - symmetry of space - time , also with the temporal separation of objects .
the name `` informational resonance '' or `` he d '' is connected with this special property of arbitrarily separated objects to be `` tuned '' to each other . as it will be discussed in the next chapter and more in the second part of the paper , this does not mean that heds do not influence the spatio - temporal relations .
today the main indication of the possibility of existence of interactions of this kind is the quantum entanglement . for concreteness consider the example of
two separated particles with spin one half , in the singlet state .
measuring spin of one of the particles along some direction in space , the second particle instantaneously ( at least , on the level of the formalism ) starts to have spin opposite to the measured one . following literal interpretation of the formalism , it can be assumed that a real interaction of two particles happened instantaneously on the level of pure information , in this case the quantum one . because the distance is unimportant here ( quantum entanglement does not decay with the distance ) then ,
considering the viewpoint of the moving observer with the account of the lorentz symmetry of space - time , we conclude that interactions of this kind may involve objects that are arbitrarily separated from each other both in space and in time .
it can be said that the objects are in an informational connection independent of the spatio - temporal separation .
speaking from a more general viewpoint , the quantum mechanics unambiguously indicates the possibility that behind it there is a theory with non - local interactions .
objects that resonate `` information - wise '' are in the state of knowledge of each other .
we use the term `` knowledge '' in a concrete meaning : for example , two entangled particles have a certain type of knowledge of each other because whatever happens with one of the particles , the second one immediately `` gets to know '' about this , being in a hard connection with the second particle .
below we will assume that the quantum entanglement corresponds to a certain he d of objects and we will use its known properties to describe properties of heds that appear general . in the frame of this assumption a most important , fundamental property of purely - informational interactions becomes clear , namely their indivisibility - a complex system participates in a he d , as a whole .
in the example of entanglement considered above this property can be seen considering systems instead of particles .
he d involves complex , spatially extended systems instantaneously and completely .
the assumption that complex systems participate in heds principally as unified , indivisible objects is in agreement with the characteristic features of the phenomenological description of nature . in different natural situations particular `` elementary particles '' -
`` indivisible objects '' arise , pertinent to the considered situations .
these `` particles '' are by themselves complex and consist of more elementary units organized into a unified whole to form that particle .
we will call such an `` elementary '' particle a `` creit '' ( from `` create '' and `` crate '' ) .
creit is a complex object that can be considered structurally stable in the considered situations , where its `` abilities '' to participate in heds may be considered constant ( properties of creits are considered in more detail in sections vi - vii ) .
let us add that wholeness of heds leads to the necessity to consider various situations on the whole , well known in the quantum mechanics .
thus , in the heds - web approach , the spatio - temporal relations are considered not as an inherent part of nature , but just as a way of measuring the latter .
in contrast , the objects interactions are considered as an inherent part of nature . at the same time , in many situations it is possible to define the spatio - temporal relations in a consistent way and find out that they obey approximately the connections described by the laws of the classical mechanics . in those special cases where there is no chaos in the system , the approximate character of the laws of mechanics is not so important and
the mechanical connections play a dominating part in the connections of the system objects .
in particular , this is the situation of two point objects with a central force interaction .
this situation , to a large extent , is at the origin of the notion that in the interaction of two objects the spatio - temporal relations are a closed , self - explaining entity that does not allow for any additional influence besides the laws of mechanics . in reality , however , the objects are usually complex and besides special cases the system is chaotic and the laws of mechanics bring little information .
it is in a chaotic situation that heds play a decisive role .
their influence is realized via small corrections to the laws of classical mechanics that arise due to the incompleteness of the spatio - temporal section of reality , see sec .
[ chapthirteen ] .
these small corrections , being amplified by chaos , lead to a significant global change and , in the end , due to their stable informational direction , it is them who fix the resultant global structure of chaos . the latter may differ radically from the one expected on the basis of the locally applicable classical mechanics .
the arising picture of chaos is a complex order where the global structure is determined by quite stable influences of heds of different types .
such a picture differs strongly for the traditional one where chaos is considered as disorder , while organization in chaos appears to be largely phenomenological .
it should be stressed that the order of heds is a complex order which is no simpler than the order described by the quantum mechanics .
consider as an example a he d of two creits .
resonating creits are in tune with each other , though the logic of this tuning may be very complex . in particular , from the viewpoint of a classical observer ( observer in space - time ) the objects behavior may look random
. nevertheless , the creits are in informational connection so that `` randomness '' has its logic - the logic of the informational resonance .
the latter leads to hardly identifiable patterns and tendencies in chaos , subtle correlations and a kind of coordination in the behavior , in short - introduces non - randomness in chaos .
this creates the principal possibility of a statistical description akin to the one that emerged in the quantum mechanics .
we pass to the consistent construction of the theory .
the heds - web approach is a purely informational approach to physics that aims to construct physics starting from purely informational concepts of objects and their informational connections - heds .
we suggest that from the viewpoint of pure information , an object is an entity possessing information like in the ordinary view an object is an entity possessing energy .
thus an object ( `` creit '' ) is essentially , a certain `` crate '' of pure information .
creits create heds - webs by entering into heds . within the heds - web approach
, the world constitutes a giant resonant web - a set of objects unified by purely informational connections into a single general structure .
this web is in the state of a constant flux - an infinite , incessant change of connections , happening everywhere and in all directions ( in the sense of flux , this state is most close to the state of quantum fields , however it is not random and it does not happen in space and time .
thus a new category is needed ) . any attempt to `` stop '' this flux and `` seize '' the state of the web , thus decomposing
the flux into the `` current state '' and its changes introduces an inaccuracy , both qualitative and quantitative . besides
, a linear sequence of causes and effects , prescribing a certain direction of the flux , is inapplicable for the heds - web ( remind that heds are non - local in time ) . in this paper
we will consider an approximation within which one can conditionally single out `` pieces '' of the web and speak of the structure of these pieces and the processes of their change .
let us note that speaking of these processes it makes no sense to speak of their temporal duration - time is a way of measurement connected with matter and it does not apply to the processes occurring at the level of pure information .
if it is necessary to perform a fundamental consideration of the structure of reality , one may choose as elementary constituents of the web sufficiently stable elementary particles , such as those appearing in the standard model or more fundamental .
the determination of stable constituents presents no difficulty in the practically important situations , that are of main interest to us here . in the heds - web approach
the spatio - temporal relations between objects are considered as metric characteristics of the web that are resultant from heds . in particular , the interactions of objects in space and time that are nothing more than the approximate description of the logic of changes of spatio - temporal relations between objects , are also considered as resultant from heds .
the formalism described below is based on the maximally economic , in our view , assumptions on the structure of the web and the flux that include the following propositions : -connections can be of different types - tones -connections can be more strong or less strong -entering into he d is a threshold phenomenon ( this is necessary to define a non - trivial dynamics , otherwise all creits will immediately enter in heds and by this the dynamics will end .
the threshold nature of heds is also important for the emergence of a natural mathematical structure to describe transformations of pure information in heds , see below ) -a web formed by resonating creits is able to act in the world of pure information as a whole .
this is demanded by consistency because creit is a complex object and thus is also a web , see fig @xmath0 .
the web properties as a whole are a non - trivial function of the informational characteristics of the creits comprising the web . [
cols="^,^ " , ] while the detailed consideration of the emerging web structures is beyond the frame of this article , their general character can be realized empirically as follows .
it is a universal property of nature to produce separation of `` scales of information '' , leading to the existence of approximately closed systems of objects that interact among themselves much more intensely than with the environment . in the frame of our approach
such picture means that every `` macro - piece '' of the web consists of many intertwined , interdependent webs , where a conditional sub - web may be singled out by separating the scales of information ( we consider the strengths of the connection between creits , as measures of their proximity , cf .
@xmath1 ) .
one can imagine a structure similar to the structure of the visible universe where stars group into galaxies , while galaxies into clusters of galaxies .
then from the viewpoint of the scale of the galaxy , the latter is a web , while from the viewpoint of the scale of the cluster - the galaxy behaves approximately as a unified whole or an effective creit .
we expect creits to have their own `` informational strategy '' or directionality of the informational connections into which they enter .
this assumption is connected with the expectation that the extreme complexity of the heds - web described above should lead to great sensitivity of the the web to the properties of the creit , that is one can expect to have a rather robust correspondence between the creit properties and its connections .
the quantitative expression of this expectation is `` the law of the dominant '' . out of the many connections in which a creit finds itself one can distinguish the dominant one , - the connection which strength is maximal .
the dominant connection significantly determines the rest of the creit connections because other connections exist on the background of a more information - saturated field - space created by the dominant connection .
this can be seen by a thought experiment considering how the existing web of the creit builds up starting from the dominant connection and continuing with the rest of the connections .
`` the law of the dominant '' says that the dominant connection introduces an essential orientation in all the rest of the connections of the considered creit .
the importance of the dominant is that it is the dominant that determines what could be called `` the informational strategy '' .
the rest of the connections are relatively subordinated to this strategy , that is to the dominant connection .
the dynamics of the heds - web is centered at the assumption that the heds - web is in the state of an infinite flux of a constant change of connections happening everywhere and at every `` point '' in all `` directions '' .
this incessant change of connections is the essence of the phenomenon we call chaos ( let us stress that in contrast to the traditional view , here chaos is the dynamics of the connections and not of objects ) . the basic proposition regarding the processes of changes of connections is that all heds that are allowed by the present thresholds may happen . in particular , creits may enter an indefinite number of connections .
the rearrangement of the connections leads to a redistribution of the pure information in the web .
this redistribution occurs constantly .
one can expect the existence of heds of three different dynamical types ( in contrast to static types - tones ) that describe different dynamical possibilities . at a `` constant '' he d of two creits
they are all the time in the he d .
a single - time he d corresponds to the situation where creits enter into he d , that after producing its effects ends and the creits do not enter in heds any more .
finally , a repeating ( pulsating ) he d corresponds to the situation of repeating acts of entering into the he d and ending it . in the heds - web approach chaos
is not a disorder , but to the contrary it is an extremely complex and definite order of changes of connections in the heds - web ( `` the world is chaotic but not arbitrary or accidental '' ) .
this order is significantly determined by `` the law of changes of dominants '' that is the dominant connections . because it is the dominant connection that orients all connections of a creit , then significant rearrangements of the web occur just at the changes of the dominants ( entering into a connection stronger than all the existing ones ) .
each such change is a `` butterfly effect '' that is a change of one connection in the web that leads to changes of many other connections .
let us note that in the heds - web approach due to the presence of definite elements - connections , the notion of the `` butterfly effect '' has a definite meaning .
in the frame of our hypothesis the spatio - temporal relations constitute a certain description or an observable section of the heds - web .
these relations are a certain manifestation of full informational connections between the creits of the heds - web .
we first consider the spatial description of the heds - web . the existing heds - web , consisting of creits and connections between them , leads to the existence of a certain spatial configuration of objects in the classical order known to us .
the spatial distance between two given objects is determined not only by the connections of the corresponding creits but also by the whole picture of connections in the heds - web .
it appears sensible to assume the existence of a positive correlation between the strength of the creits connection and their spatial proximity , manifested depending on the whole set of connections in the heds - web .
a convenient way for describing the correspondence between the heds - web and the spatial configuration of objects is the consideration of the correspondence between the distribution of information in the field - space and the distribution of the mass - energy in space .
the latter constitutes quite a complete description of the spatial configuration where objects are represented by a sharp increase of the mass - energy density .
analogously , the distribution of information characterizes creits and their closeness in the sense of the strength of their connection . in accord with the above
we introduce the first postulate of the correspondence .
correspondence between field - space and space ( the latter understood as a space - like hypersurface in space - time ) : the distribution of information in the field - space of the heds - web is manifested in space as a certain distribution of energy .
thus , the mass - energy density @xmath2 appearing in the einstein equations reflects the distribution of information .
let us remind that our general approach is aimed at considering what stands behind space and time so to achieve a more complete view of the latter and the laws described within their frame .
correspondingly we will assume that the space - time emerging from the heds - web is described by the general theory of relativity , that is by a metric satisfying einstein s equations .
this leads to the second postulate connecting the distribution of energy , reflecting information , and the space - time geometry .
correspondence between the energy distribution on space - like hypersurfaces and the space - time metric : this correspondence is based on the demand that the cartan moment of rotation and the density of mass - energy are equal for each space - like hypersurface according to @xmath3 where @xmath4 is the scalar curvature invariant of the @xmath5geometry intrinsic to the hypersurface , while @xmath6 is the extrinsic curvature tensor .
here and below we follow the notations of @xcite .
as it is well known the second postulate leads to the einstein equations @xmath7 where @xmath8 is the einstein tensor and @xmath9 is the stress - energy tensor .
the correspondence is realized as follows .
designating the local @xmath10vector normal to an arbitrary spacelike slice through spacetime by @xmath11 , one has @xmath12 then eq . ( [ basic1 ] ) leads to @xmath13 which is equivalent to eq .
( [ basic2 ] ) by the arbitrariness of the spacelike slice .
the main new component of the suggested approach to the space - time is the outlook at the energy distribution as described by the postulate 1 .
the questions of what the energy is and how it is connected with the notion of information are usually outside the frame of physics . in the suggested approach the mass - energy distribution in space reflects the distribution of information in the field - space created by the creits and their connections .
( it should be added that the mass - energy density above becomes meaningful only after the metric is found . )
even though the second postulate is one of known ways of establishing a connection between the energy distribution and the metric , it acquires a new qualitative meaning from the viewpoint of heds .
because the mass - energy density reflects the distribution of information created by creits and their heds , then eq . ( [ basic1 ] ) says that creation of additional heds between given creits leads to additional curving of the space - time .
for example , when two creits enter into he d some additional information appears ( because of the super - additivity ) that leads to the change of the energy distribution and eventually the cartan moment of rotation .
thus , heds influence what happens in space - time by changing the moment of rotation ( the corresponding influence on chaos is discussed in the following two sections ) .
postulate 1 also gets an additional meaning . because every creit in the heds - web , generally speaking ,
participates in many connections , then it can be said that the heds - web possesses a certain generalized `` elasticity '' .
the existing connections `` resist '' the creation of new connections in accord with the law of the change of the dominant .
then the parallel between the theory of curved space - time and the elasticity theory , pointed out by sakharov @xcite , acquires a new meaning - the einstein equations express the metric aspect of the heds - web elasticity .
let us note however that the parallel with the sakharov view , who spoke of elasticity of space , is but qualitative - in our approach the space does not exist by itself and the heds - web is not embedded in space in contrast to the quantum vacuum fluctuations .
also the heds - web does not constitute a `` skeleton '' for space of a kind discussed in the regge calculus @xcite ( rather it a `` skeleton '' for the field - space ) .
the bianchi identity @xmath14 allows to obtain from the einstein equations ( [ basic2 ] ) the law of energy - momentum conservation @xcite @xmath15 thus , in the heds - web approach it appears natural to consider the law of energy - momentum conservation as a constraint resultant from the use of the spatio - temporal coordinates to describe pure information .
it is known that @xcite the laws of motion of the classical physics can be obtained from eq .
( [ basic4 ] ) by prescribing a correct constitutive relation for the tensor @xmath16 .
for example , the maxwell equations can be obtained by substituting into eq .
( [ basic4 ] ) the constitutive relation @xmath17 where @xmath18 is the metric and @xmath19 is determined by the @xmath10potential @xmath20 via @xmath21 .
classical particle dynamics may also be obtained in this way @xcite ) we see that the above derivation of the maxwell equations does not differ in essence from the derivation of the equations of hydrodynamics where the energy - momentum conservation is considered as given and then a constitutive relation for @xmath16 is prescribed ( the particle number conservation is also added as a rule ) . in particular , the field @xmath20 plays the role of a `` slow '' field or the order parameter . as it is well known in hydrodynamics , the used expressions for @xmath16
are effective approximate relations . in accordance with this
, it seems natural to consider the possibility that the laws of physics are effective , empirically confirmed descriptions of the stress - energy tensor , that provide for a well - working locally in space and time approximation to the dynamics of energy .
that is , guided by the spatio - temporal logic , good , self - consistent local approximations for @xmath16 are achieved in terms of some effective variables which determination is the subject of the research .
it is these approximations that constitute the laws of physics
. such a non - absolutist view may be very useful for approaching such problems of the usual formal approach as for example the `` infinite self - energy '' of a point particle ( the decomposition of @xmath16 into the field and the particle becomes the question of finding an effectively working description ) .
it is more important to us here that this demonstrates the existence of corrections to the known expressions for @xmath16 and thus to the equations of motion .
let us introduce the decomposition @xmath22 where @xmath23 is the component of the tensor the expression of which is known from the laws of physics , while @xmath24 is the correction existing due to the existence of the more fundamental description .
formally this component can be defined via ( let us note the similarity of the used procedure with the one used in the analysis of the dark energy , see e. g. @xcite ) @xmath25 then the usual laws of motion that would follow from @xmath26 acquire two corrections - one due to the correction to the metric produced by the term @xmath24 in the einstein equations , while the other is due to the appearance of an additional term in the rhs of the equation @xmath27 [ here it makes no sense to analyze where one can shorten the equation to @xmath28 . in the case where the solutions of the original equations following from @xmath26 are sensitive to the corrections , the resulting picture obtained with the account of
the latter may differ qualitatively from the one expected on the basis of the `` spatio - temporal '' approximation @xmath29 .
according to our assumption this is just the case of the chaos .
thus , according to the heds - web hypothesis , the equations of the classical mechanics must have small corrections due to @xmath24 ( not only of quantum nature ) . from the viewpoint of the spatio - temporal section , these corrections are illogical .
the apparent absence of logic in them is determined by the discrepancy of the spatio - temporal description and the true logic of chaos - the logic of heds . from the viewpoint of a classical observer
, it would be natural to model these corrections as random which is just what is effectively done in chaos where a statistical description is introduced almost necessarily .
in particular , such explanation of the effective randomness could lead to further understanding of the foundations of the statistical mechanics and of the reason why the introduction of noise in the equations is so effective .
a classical observer sees the spatio - temporal results of the heds without seeing their cause .
these results obey approximately the laws of mechanics which are then considered to be their causes . according to the heds - web hypothesis the mechanics in its approximations misses the essence of chaos as a complex order of the heds - web .
the amplification of the locally small correction due to @xmath30 , that distinguishes the heds - web order from the mechanical disorder , leads to a qualitative difference of the global picture from the local one .
in fact , as it was shown in the previous section , the laws of mechanics ( as the laws relying on the energy - momentum conservation ) constitute rather a geometrical limitation due to the choice of the way to consider chaos , than the law of chaos .
it is this fact that in our view stands behind the fact that mechanics brings little information when chaos is considered . in other words ,
the laws of mechanics look like constraints on the manifestation of heds related to the introduction of the spatio - temporal system of coordinates .
this is also the reason for the big difference of notions of a mechanical and a resonant connections - one means a constraint , the other means increase of possibilities .
heds effect on the space - time is indirect - the cartan moment of rotation characterizing the spacelike slices is their basic manifestation . the importance of the cartan moment of rotation for the laws of physics was stressed in @xcite . from the view - point of space and time heds are manifested via contributions to the energy - momentum tensor that do not allow for an effective spatio - temporal description ( forms of energy not allowing an effective embedding in the space - time ) .
this clearly reminds of the dark energy the consideration of which , however , is beyond the frame of this work .
the emerging outlook at randomness in the quantum mechanics is quite similar to the outlook at randomness in the classical chaos with a sole exception that quantum mechanical randomness is considered conventionally as fundamental . from our viewpoint
this fundamentality corresponds to the principal inapplicability of the spatio - temporal order to the consideration of quantum connections ( these connections may manifest heds independently of the spatio - temporal separations as in the example of the quantum entanglement ) .
the statistics is a description of the interaction of a classical observer , acting within the spatio - temporal logic , with a quantum object participating in connections which character is not spatio - temporal in principle , taking account for the difference of scales of interacting objects .
let us stress that in the heds - web approach there are no classical and quantum objects , there are only creits participating in that or another connection .
the heds - web is a possible objective reality behind the quantum mechanics .
the study of the possibility to follow in detail the emergence of a quantum mechanical description is a subject for future work .
we considered a possible continuation of the reconsideration of the notions of space and time started by the general theory of relativity .
the demand not to consider the space - time geometry among the primary elements of the theory was extended to the demand not to consider among them the space - time as such .
this approach is a natural one for the search of a more fundamental theory that could unite the general theory of relativity and the quantum mechanics .
we suggested that the theory standing behind the space and time is purely informational .
as a structure of such a theory we suggested a web - `` the heds - web '' .
the latter is formed by creits ( analogs of objects in the `` information space '' - field - space ) and their connections - `` informational resonances '' or `` heds '' .
the most important consequence of our consideration is a new outlook at the classical mechanics that reveals the purely informational character of the order of chaos .
this order can not in principle be caught within the frame of the existing approach based on space and time . in the paper
the foundations of the heds - web approach were established and the discussion of laws of transformations of pure information when creits enter into heds , was started .
the notion of the field - space , as a unified information field of the heds - web , was introduced , and the problem of introducing a new category , connected with the infinite flux state of the heds - web , was posed . within the frame of the heds - web approach , chaos is a definite order of changes of connections in the heds - web , that is chaos in no way means an accident or an arbitrariness ( this claim encompasses the quantum mechanics too ) .
the main suggested law of the order of chaos is the law of changes of the dominants - the strongest connections of the creits .
the changes of the dominants occurring due to the appearance of another , stronger dominant connection rearrange many connections at once , constituting the `` butterfly effect '' of the heds - web .
we gave a non - contradictory view at the emergence of the laws of classical physics from the pure information .
in this view the energy distribution in space reflects the distribution of information in the field - space .
the laws of physics emerge as effective descriptions of the stress - energy tensor , similar in nature to constitutive relations for the stress - energy tensor used in the hydrodynamic approach . while constituting a good approximation locally in space and time
, these laws may give an incorrect global picture where chaos is present .
this is connected with the fact that the approximation used in mechanics is too crude to capture completely the informational connections in chaos determining the global picture . in other words
, there is a qualitative difference between the local mechanical approximation to the laws of chaos and the informational nature of the latter . as a result
, mechanics often gives just a self - consistent view that , not explaining anything , describes the result occurring in nature de facto .
the heds - web describes the inner structure of chaos as a constant change of connections of objects , but not of objects themselves , that remain unchanged .
formally both the heds - web and the classical mechanics say that chaos is determined .
however , the classical mechanics is not closed because of the presence of the more fundamental quantum - mechanical description .
the statistical nature of the latter eventually brings the conclusion that within the traditional approach , based on space and time , chaos is fundamentally random .
the heds - web approach , by disclosing the structure standing behind the classical and the quantum mechanics , restores the non - randomness of chaos .
the determinism , however , reemerges at a new level - this is already not determinism in time , but determinism in connections . suggesting a structure of the objective reality standing behind both the quantum mechanics and the space and time , the heds - web is a candidate for the theory of quantum gravity . from the viewpoint of the quantum mechanics , heds
play the role of the `` hidden '' non - local interactions .
being informational connections akin to the quantum entanglement , heds correspond to informational dynamics not simpler than the quantum - mechanical one . at the same time , both , micro- and macro- objects may participate in heds .
that is , within the frame of the suggested picture of the objective reality , macro - objects may enter into connections not less complicated than the quantum - mechanical ones .
it is these connections , applicable at all scales , that constitute the principles of the organization of chaos .
thus , if the heds - web hypothesis is correct , then one may expect phenomena not less surprising than the quantum - mechanical ones in the physics of `` ordinary '' macroscopic objects .
in other words , the heds - web hypothesis points out to the possibility of principally new ways of operating objects .
in particular , it is not excluded that the principles described in this paper can be used to achieve high - temperature superconductivity .
the necessity of the statistical description in chaos ( where usually statistics is considered non - fundamental ) and in the quantum mechanics , can now be viewed as a consequence of the incompleteness of the spatio - temporal description of nature .
the properties of complex systems , as a whole , are probably the main characteristics of the organization of chaos that is absent in the traditional fundamental approach , where they appear as emerging essentially accidentally .
usually they do not contradict the known laws of interactions of parts , but also can not be obtained from them formally . in the heds - web approach ,
the complex systems , constituting resonant webs , possess a larger amount of information than just the sum of the amounts of information of the creits comprising the web .
thus , in this approach , the emergent properties of complex system as wholes are fundamental . in view of the fact that the emergence of properties of complex systems as wholes is an overwhelmingly widespread phenomenon , the outlook that the emergence is a fundamental property of nature appears to us more satisfactory .
heds , that according to our assumption are as fundamental interactions of objects as the known ones could also shed light at the fundamental questions of physics .
in particular , the instantaneous nature of heds could help to resolve some principal difficulties of the modern fundamental theories of matter . in the latter
, the assumption of a finite maximal speed of propagation of information leads to the impossibility of considering spatially extended objects as the elementary ones . as a result
, one has to consider either point particles or fields determined over continuum that brings about different kinds of divergences .
summarizing , the reconsideration of the notions of space and time brings about two principally new possibilities that could be seen while remaining within the frame of the spatio - temporal approach .
first , there appears a possibility that chaos has an informational nature .
second , there appears a possibility of principally new ways of operating objects . in connection with this
, we suppose that the possibility of the passage to the purely informational physics , suggested in this work , is of high interest .
|
we study the possibility that further advancement in the understanding of the order of chaos may demand a certain reconsideration of the approach to the classical mechanics . for this
we suggest to consider the viewpoint that spatio - temporal relations between objects are emergent and that they are but a result of complex interactions of objects .
such an approach is a natural continuation of the revision of the notions of space and time started by the general theory of relativity .
it leads to a possible extension of the structure of the classical mechanics .
namely , the result of the objects interactions can be wider than the spatio - temporal relations . in this case interactions
form a generalized space ( `` field - space '' ) wider than its spatio - temporal section ( while the known interactions are embedded within that section ) .
the study of such hypothesis demands constructing a theory , not relying on space and time as primitive notions . as primary elements of such a theory
we consider objects and purely informational connections between them , not expressed in spatio - temporal terms . following the logic of this theory
, the world constitutes a complex information web .
the web is in the state of a constant flux ( of a more general category than the one of the quantum fields ) - an incessant change of connections , the complex order of which comprises the order of chaos . reminding the space build - up from the regge skeleton ,
the web builds a unified field - space of pure information , manifested in space as energy distribution .
we describe the emergence of spatio - temporal laws from the viewpoint of purely informational physics .
the suggested construction of physics on the purely informational basis is a candidate for the theory of quantum gravity .
|
quantum coherence is a phenomenon that could have profound impact on technological development .
usually understood to appear in temperature regimes much lower than those encountered in everyday practice , quantum coherence effects are not directly utilized in the operation of most conventional devices . superconducting sensors , magnets and transmission cables are examples where macroscopic quantum coherence is maintained at relatively high temperature .
nonetheless , even the highest-@xmath0 materials currently remain in the realm of cryogenic temperatures , thus significantly hampering their potential applicability .
consequently , much effort has been expended in the quest for higher temperature materials , as room - temperature superconductivity remains one of the most prominent issues in condensed matter physics .
although recent experimental evidence may suggest that superconductivity is possible at room temperature @xcite , what is perhaps even more compelling is the emerging evidence that photosynthetic systems found in nature may have already evolved a means of exploiting quantum coherence effects at physiological temperatures @xcite . due to the broad array of organisms sustained by trapping
then converting sunlight into chemical energy , photosynthesis is probably the cleanest , most efficient energy collection mechanism on earth , perhaps making high - temperature quantum coherence far more ubiquitous than previously thought .
central to the understanding of temperature - dependent quantum coherence is the decoherence or thermalization of quantum systems with their surrounding environment . in the well - known work of caldeira and leggett @xcite , the feynman - vernon influence functional approach @xcite
is used to model a thermalizing environment with the purpose of introducing decoherence into a quantum oscillator . following the same path integral approach , it is demonstrated that a thermalizing environment need not always lead to decoherence in the steady - state .
arriving at this result requires the realization that because the reservoir can not excite energies too far in excess of its mean thermal energy , a temperature cutoff must be introduced . as a result , low - temperature environments are found to thermalize the quantum oscillator to the usual boltzmann distribution of energy levels , whereas higher - temperature regimes induce a single , coherent oscillator ground state with an imaginary frequency .
since the concomitant imaginary energy eigenvalues correspond to decay rates , all excited states become more unstable with increasing quantum number . in this way
, environments in this higher - temperature regime damp out the excitations by extracting energy from the quantum oscillator , forcing the oscillator to remain in the stable , coherent ground state . increasing to very high environmental temperatures once again
thermalizes the quantum oscillator in the usual way
. in short , a dissipative environment can cause a suppression or damping of excited modes , thus acting to maintain quantum coherence even at relatively high temperatures .
this result may underpin recent observations of room - temperature superconducting effects in water - treated graphite grains @xcite .
finally , given their emergence and subsequent disappearance of quantum coherence with increasing temperature , photosynthetic molecules most remarkably follow the qualitative behavior described here @xcite .
depicted schematically in fig.[qq ] , a quantum harmonic oscillator , initially at some temperature @xmath1 , linearly coupled to an environmental reservoir .
ultimately the environment will be modeled as a collection of oscillators at temperature @xmath2 , but we first consider the case of a single - oscillator reservoir , carrying out integrals over a continuum later . acquiring the steady - state of the quantum harmonic oscillator
requires the time evolution of its corresponding density matrix .
[ h ] = 3.25 in oscillator @xmath3 occupies the usual boltzmann distribution of eigenstates associated with temperature @xmath1 . because each state occurs with a relative probability of @xmath4 ,
the initial density matrix is given by @xmath5 = \int \ !
\rho_0(q_i , q_i ) \
, dq_i , \label{z01 } \end{aligned}\ ] ] where @xmath6 are the unperturbed eigenfunctions of oscillator @xmath3 , and @xmath7 is shorthand for the initial point @xmath8 of the oscillator trajectory . given a harmonic oscillator of mass @xmath9 and frequency @xmath10 , expression can be summed explicitly to give @xmath11 \vphantom { \frac{m \ , \omega_0}{2 \hbar \sinh ( \beta_0 \hbar \omega_0 ) } } \ !
\right\ } , \label{denmat0}\end{aligned}\ ] ] and z_0 = .
[ z02 ] throughout , the parameters @xmath12 and @xmath13 assume the usual meanings as the respective inverse temperatures , [ beta0 ] = , _ 0 = . alternatively , the initial state of the oscillator @xmath3 may be taken to be one of the eigenstates @xmath6 so that _ 0(q_i , q_i ) = _ n(q_i ) _
n^(q_i ) , z_0 = 1 . in either case
, the time evolution of the density matrix is given by [ timeev ] ( q_f , q_f,t ) = e^ f(q , q,t ) _ 0(q_i , q_i ) dq dq dq_i dq_i . just as @xmath7 refers to the initial trajectory point @xmath8 ,
@xmath14 is shorthand for the final point @xmath15 .
furthermore , the density matrix of the quantum oscillator is expressed as a path integral in which the coordinates of the environmental oscillator @xmath16 have been integrated out . as a result
, there emerges an influence functional , @xmath17 , which describes the effect of the environmental oscillator solely in terms of the quantum oscillator coordinates .
effectively , the coupled problem has been reduced to the dynamics of an isolated oscillator weighted by the kernel @xmath17 .
further details may be found in @xcite and @xcite .
for the case of a thermal reservoir modeled by a harmonic oscillator , the influence functional has the form @xmath18
\int \limits_0^{t ' } \left [ q(t '' ) \alpha ( t'-t '' ) \right .
\nonumber \\ &
\qquad \qquad \qquad \qquad \qquad \qquad \qquad \left
. - q'(t '' ) \alpha^\ast ( t'-t '' ) \right ] dt '' \ , dt ' \vphantom{- \int \limits_0^t } \biggr\ } , \end{aligned}\ ] ] where ( t-t ) = .
these expressions for @xmath17 and @xmath19 are well - known since they form the basis of the caldeira - leggett model @xcite , having been directly obtained from sec.12 - 9 of @xcite .
consider the first integral in the exponential of @xmath20 , @xmath21 dt''$ ] . because the environment is modeled by a continuum of oscillators
, we must sum this expression over all frequencies @xmath22 .
the contribution from the j_th _ oscillator is @xmath23 \ !
dt '' = \nonumber \\ & \qquad \qquad \qquad \frac{c_j^2}{2m_j \omega_j \hbar } \frac{1}{e^{\beta \hbar \omega_j } - 1 } \int \limits_0^{t ' } \left\ { q(t '' ) \ !
e^{i \omega_j ( t'-t '' ) } + e^{\beta \hbar \omega_j } e^{-i \omega_j ( t'-t '' ) } \right ] \right .
\nonumber \\ & \qquad \qquad \qquad \qquad \qquad \qquad \qquad \left . -q'(t '' ) \ !
e^{-i \omega_j ( t'-t '' ) } + e^{\beta \hbar \omega_j } e^{i \omega_j ( t'-t '' ) } \right ] \right\ } \ !
\end{aligned}\ ] ] defining the spectral density as [ jdef ] j ( ) = _ j ( - _ j ) , the sum over all oscillators can be written as an integral @xmath24 \ !
dt '' \nonumber \\ & \qquad = \frac{1}{\pi \hbar } \int \limits_0^\infty \frac{j(\omega)}{e^{\beta \hbar \omega } - 1 } \int \limits_0^{t ' } \left\ { q(t '' ) \
! \left [ e^{i \omega ( t'-t '' ) } + e^{\beta \hbar \omega } e^{-i \omega ( t'-t '' ) } \right ] \right .
\nonumber \\ & \qquad \qquad \qquad \qquad \quad \left . -q'(t '' ) \ !
e^{-i \omega ( t'-t '' ) } + e^{\beta \hbar \omega } e^{i \omega ( t'-t '' ) } \right ] \right\ } dt '' d \omega\ , .
\end{aligned}\ ] ] it is convenient to define [ adef1 ] a(t-t ) = _ 0^ d . assuming the spectral density , @xmath25 , to be real , the real part of @xmath26 is [ ardef1 ] re\ { a(t-t ) } = a_r ( t-t ) & = _ 0^ ( ) j ( )
d + & = a_r(t-t ) . similarly ,
the imaginary part of @xmath26 is [ aidef1 ] i m \
{ a(t-t ) } = a_i(t-t ) = - _ 0^j ( ) d . using the definition of @xmath26 , the sum over all oscillators
simplifies to @xmath24 \ !
dt '' \nonumber \\ & = \int \limits_0^{t ' } \left\ { a_r(t'-t '' ) \left [ q(t '' ) - q'(t '' ) \right ] + i a_i(t'-t '' ) \left [ q(t '' ) + q'(t '' ) \right ] \right\ } dt ''
. \end{aligned}\ ] ] substituting this expression into the influence functional yields @xmath27 \int \limits_0^t a_r(t'-t '' ) \left [ q(t '' ) - q'(t '' ) \right ] dt '' dt ' \nonumber \\ & \hphantom{\exp \left\ { \right\ } } -i \int \limits_0^t \left [ q(t ' ) - q'(t ' ) \right ] \int \limits_0^{t ' } a_i ( t'-t '' ) \left [ q(t '' ) + q'(t '' ) \right ] dt '' dt ' \biggr\ } .
\nonumber \\[-.05in]\end{aligned}\ ] ] recall that the action for the quantum oscillator of interest , @xmath3 , is given by s(q ) = m _ 0^t dt . given this action in addition to result obtained for the influence functional ,
the time evolution of the density matrix is @xmath28 \int \limits_0^t a_r(t'-t '' ) \left [ q(t '' ) - q'(t '' ) \right ] dt '' \nonumber \\ & \qquad \left .
- \frac{2 \hbar}{m } \left [ q(t ' ) - q'(t ' ) \right ] \int \limits_0^{t ' } a_i(t'-t '' ) \left [ q(t '' ) + q'(t '' ) \right ] dt '' \biggr\ } dt ' \right ) \nonumber \\ & \qquad \qquad \cdot \rho_0 ( q_i , q'_i ) dq \ , dq ' \ , dq_i \ , dq'_i .\end{aligned}\ ] ] transforming to the rotated coordinates , [ xydef ] @xmath29 \label{xdef } \\[.1 in ] y(t ' ) & = \frac{1}{\sqrt{2 } } \left [ q(t ' ) - q'(t ' ) \right ] , \label{ydef } \end{aligned}\ ] ] may be written more succinctly as @xmath30 dt ' \right\ } \\ & \qquad \qquad \qquad \qquad \qquad \qquad \cdot \rho_0 ( x_i , y_i ) dx \ , dy \ , dx_i \ , dy_i .
\end{split}\end{aligned}\ ] ] as indicated by the resulting expression , the density matrix gives rise to a single , composite action @xmath31 & = m\int \limits_0^t \biggl [ \dot{x}(t ' ) \dot{y}(t ' ) -\omega_0^{\,2 } \ ,
x(t ' ) y(t ' ) \nonumber \\ & \qquad \quad + i \frac{\hbar}{m } \ , y(t ' ) \ !
\int \limits_0^t a_r(t'-t '' ) y(t '' ) \ , dt '' \\ & \qquad \quad - \frac{2 \hbar}{m } \ , y(t ' ) \ !
\int \limits_0^{t ' } a_i(t'-t '' ) x(t '' ) \ , dt '' \biggr ] dt ' .
\nonumber \end{aligned}\ ] ] let @xmath32 be the classical trajectory between the fixed initial point @xmath33 and the fixed final point @xmath34 , and let @xmath35 be the classical trajectory between the fixed initial point @xmath36 and the fixed final point @xmath37 . by definition , @xmath32 and @xmath35 must extremize the action @xmath38 . in the usual way ,
@xmath39 and @xmath40 are expressed as the sum of their corresponding extremal paths plus their respective fluctuations , @xmath41 and @xmath42 : @xmath43 @xmath44 in terms of @xmath32 , @xmath41 and @xmath35 , @xmath42 , the composite action becomes @xmath45 = \nonumber \\ & m\int \limits_0^t \biggl [ \dot{x}_{cl}(t ' ) \dot{y}_{cl}(t ' ) - \ddot{x}_{cl}(t ' ) \delta y(t ' ) - \ddot{y}_{cl } ( t ' ) \delta x(t ' ) + \delta \dot{x}(t ' ) \delta \dot{y}(t ' ) - \omega_0 ^ 2 \ , x_{cl}(t')y_{cl}(t ' ) \nonumber \\ & \qquad \qquad \qquad \qquad \qquad \quad - \omega_0 ^ 2 \ , x_{cl}(t ' ) \delta y(t ' ) - \omega_0 ^ 2 \ , y_{cl}(t ' ) \delta x(t ' ) - \omega_0 ^ 2 \ , \delta x(t ' ) \delta y(t ' ) \nonumber \\ & \quad + i \frac{\hbar}{m } \ , y_{cl}(t ' ) \ !
\int \limits_0^t a_r(t'-t '' ) \ , y_{cl}(t '' ) \ , dt '' + i \frac{2 \hbar}{m } \ , \delta y(t ' ) \ !
\int \limits_0^t a_r ( t'-t '' ) \ , y_{cl}(t '' ) \ , dt '' \nonumber \\ & \quad + i \frac{\hbar}{m } \ , \delta y(t ' ) \ !
\int \limits_0^t a_r(t'-t '' ) \ , \delta y(t '' ) \ , dt '' -\frac{2 \hbar}{m } \ , x_{cl}(t ' ) \ !
\int \limits_{t'}^t a_i(t''-t ' ) \ , y_{cl}(t '' ) \ , dt '' \nonumber \\ & \quad -\frac{2 \hbar}{m } \ , \delta y(t ' ) \ ! \int \limits_0^{t ' } a_i(t'-t '' ) \ , x_{cl}(t '' ) \
, dt '' -\frac{2 \hbar}{m } \ , \delta x(t ' ) \ !
\int \limits_{t'}^t a_i(t''-t ' ) \ , y_{cl}(t '' ) \
, dt '' \nonumber \\ & \quad -\frac{2 \hbar}{m } \ , \delta y(t ' ) \ ! \int \limits_0^{t ' } a_i(t'-t '' ) \
, \delta x(t '' ) \ , dt '' \biggr ]
dt ' .\end{aligned}\ ] ] through the rearrangement of integrals , all terms linear in @xmath46 and @xmath47 have been expressed in terms of @xmath41 and @xmath42 .
the linear terms in the fluctuations vanish , thereby constraining the classical trajectories to satisfy [ classtrajxy ] [ ycl ] _ cl ( t ) + _ 0 ^ 2 y_cl(t ) + _ t^t
a_i ( t-t ) y_cl(t ) dt = 0 , @xmath48 & \qquad \quad + \frac{2 \hbar}{m } \
! \int \limits_0^{t ' } a_i ( t'-t '' ) \ , x_{cl}(t '' ) \ , dt '' = 0 .
\end{split } \end{aligned}\ ] ] substitution of the resulting simplified action back into the time evolution of the density matrix gives [ denmatxy3 ] z_0 ( x_f , y_f , t ) = ( t ) e^ s_cl _ 0(x_i , y_i ) dx_i dy_i , where the path integral dependent prefactor is identified as @xmath49 dt ' \right\ } d\delta x d \delta y , \nonumber\end{aligned}\ ] ] and where the effective classical action is [ classaction1 ] s_cl & = m_0^t dt .
from equation for @xmath35 , we have @xmath50 , which simplifies the classical action in , [ classaction2 ] s_cl = m
. together , equations , , and give a general path integral formulation for the time evolution of a single quantum oscillator coupled to a thermal environment .
in this description , the thermal reservoir is modeled as a continuous spectrum of oscillators occupying a boltzmann distribution at temperature @xmath2 .
however , further progress requires an explicit form of @xmath51 which depends on the spectral density @xmath25 .
typically , the spectrum of the reservoir is taken to have a linear form , @xmath52 , at low frequencies .
since this spectrum can not persist to arbitrarily high frequencies , there must be a cutoff beyond which there is no appreciable coupling to the quantum oscillator , @xmath3 . because the thermal environment does not excite oscillator frequencies too far in excess of the mean thermal frequency , @xmath53
, it is reasonable to consider the case where the cutoff is twice this mean , [ j1 ] j ( ) = 2 m & || < 2 / + 0 & . whereas @xmath54 is the usual damping parameter , the inverse of the mean thermal frequency , @xmath55 , is identified as the thermalization time scale of the reservoir . using the spectral density of in the definition , the expression for @xmath56
can be written [ ai2 ] a_i(t-t ) = - _
-2/^2/ e^i ( t-t ) d. because the environmental reservoir contains many more degrees of freedom than the quantum oscillator , it is assumed that the thermalization time of the reservoir is always much shorter than @xmath57 , the thermalization time of the oscillator .
hence it is always the case that @xmath58 and the integration limits in may be extended to infinity , @xmath59 care must be exercised when taking this limit , for it implicitly contains the approximation [ a_approx ] _
-2/^2/ e^i ( t-t ) d ( t-t ) .
when appearing within an integral , this approximation permits the delta function to be used .
however , when seemingly divergent expressions like @xmath60 arise outside of any integral , it must be realized that what is really meant is [ delta0 ] ( 0 ) = .
simply substituting into the integro - differential equation gives an ambiguous result since the delta function zero occurs at one of the integration limits .
therefore , it is realized that the proper limit is obtained by half of the value of the corresponding case in which the delta function zero occurs within the integration limits .
along with the observation in , this prescription results in a second - order differential equation for @xmath35 , [ ycl2 ] _
cl(t ) - 2 _ cl(t ) + ( _ 0 ^ 2 - ) y_cl(t ) = 0 . whether the solutions to are exponential or oscillatory is determined by the discriminant [ discrim ] d = - _ 0 ^ 2 + + ^2 . for the case of extreme underdamping ,
@xmath61 , the discriminant is close to @xmath62 at low temperatures , thus giving oscillatory solutions .
however , if the temperature is sufficiently high that [ coherencecond ] > 1 , then the discriminant becomes positive , leading to purely exponential solutions of . in the infinite - temperature limit
, the term involving @xmath63 can only be convergent if @xmath64 has a temperature dependence which decays at least as rapidly as @xmath65 .
it is plausible to include these limits by taking an exponential form for the damping parameter [ epsilonmodel ] = _ 0 e^-t / t_c , _ 0 _ 0 , in which @xmath0 is a temperature cutoff .
if the cutoff is sufficiently large then the discriminant can be positive , thus giving purely exponential solutions around some neighborhood of @xmath66 . as indicated by fig.[dplot ] , this solution gives rise to a unique steady - state , to be shown through corresponding derivations of the density matrix .
[ h ] = 3.5 in
within some intermediate temperature range , the discriminant in may be positive , indicating that the general solution to the differential equation has the form y_cl(t ) = e^t \ { b ( t ) + c } , where @xmath67 and @xmath68 are constants to be determined from the boundary conditions and where we have defined [ xidef ] = > 0 . applying the boundary conditions [ bcs ]
@xmath69 gives [ yclsolnxi ] y_cl(t ) = \ { y_f e^(t-t ) ( t ) + y_i e^t } .
the required end - point derivatives are obtained from , [ ydotxi0 t ] @xmath70 \label{ydotxi0 } \\
\dot{y}_{cl}(t ) & = \gamma y_f + \frac{\xi}{\sinh ( \xi t ) } \left [ y_f\cosh(\xi t ) - y_i e^{\gamma t } \right ] .
\label{ydotxit } \end{aligned}\ ] ] in addition to the classical trajectory , @xmath35 , the action is explicitly dependent upon @xmath71 , which is arrived at by inserting the spectral density into the definition , [ ar1 ] a_r ( t-t ) = _ -2/^2/ ( ) e^i ( t-t ) d . expression is now inserted into the corresponding integrals in the classical action , @xmath72 from the solution for @xmath73 given by , the required time integrals are evaluated to be @xmath74 \right . \nonumber \\ & \qquad \qquad \qquad \quad \left .
\left . - \frac{1}{\omega + i(\xi + \gamma ) } \left [ e^{-i ( \omega + i\xi ) t } - e^{-\gamma t } \right ] \right\ } \right .
\nonumber \\ & \qquad \qquad - y_i \ !
\left\ { \frac{1}{\omega - i(\xi - \gamma ) } \left [ e^{-i ( \omega + i \gamma ) t } - e^{\xi t } \right ] \right .
\nonumber \\ & \qquad \qquad \qquad \left . -\frac{1}{\omega + i(\xi + \gamma ) } \left [ e^{-i ( \omega + i \gamma ) t } - e^{-\xi t } \right ] \right\ } \biggr ) .\end{aligned}\ ] ] after multiplying this result by its complex conjugate , only the @xmath75 integral in remains , @xmath76 \nonumber \\ & \qquad \qquad \qquad \qquad \qquad \qquad \qquad \qquad -2 y_f y_i \left [ \cosh ( \xi t ) f_\xi(\omega ) - g_\xi(\omega ) \right ] \nonumber \\[.1 in ] & \qquad \qquad \qquad \qquad \qquad \qquad \qquad \qquad + y_i^2 e^{\gamma t } \left [ f_\xi(\omega ) - e^{\xi t } g_\xi ( \omega ) \right ] \bigr\ } \nonumber \\[.1 in ] & \qquad \qquad \qquad \qquad \qquad \qquad \qquad \qquad \qquad + \ { \xi \rightarrow - \xi \ } .\end{aligned}\ ] ] for notational conciseness , the last term in curly braces , @xmath77 , is shorthand for the addition of the preceding expression but with @xmath78 replaced by @xmath79 . furthermore , the functions @xmath80 and @xmath81 are identified as [ fgdefs ] @xmath82 \bigr\ } \label{fdef } \\
g_\xi(\omega ) & = \frac{1}{\omega^2 + ( \xi - \gamma)^2 } \bigl\ { \cosh[(\xi - \gamma)t ] - \cos ( \omega t ) \bigr\ } . \label{gdef } \end{aligned}\ ] ] in this regime , the required integrals over @xmath75 have been carried out in the appendix , where it is also shown that the thermalization limit @xmath83 simplifies to [ arintsimpxi2 ] _ 0^t
y_cl(t ) _ 0^t a_r(t-t ) y_cl(t ) dt dt i ( ) ( y_f^2 + y_i^2 ) .
in addition to , the corresponding long - time limits of and are used in the classical action to obtain @xmath84 { } \ , m \ !
\left [ x_f ( \gamma + \xi ) y_f - x_i ( \gamma - \xi ) y_i -\gamma \coth \ ! \left(\frac{i
\hbar \xi}{2 } \right ) \
! \left ( y_f^2 + y_i^2 \right ) \right ] \nonumber \\ & \qquad \simeq m \xi \
! \left ( x_f y_f + x_i y_i \right ) , \end{aligned}\ ] ] where the last line follows from the condition of severe underdamping in which @xmath64 is by far the smallest frequency in the system .
consequently , the entire contribution from is negligible .
having obtained an expression for the effective action in terms of the end - point coordinates , we are now in a position to evaluate the integral @xmath85 @xmath86 occurring in the time - dependent density matrix . the initial density matrix given by eq .
must be expressed in terms of the rotated coordinates , @xmath87 \biggr\ } \ , , \end{aligned}\ ] ] where the prefactor @xmath88 is identified as [ n01 ] n_0 = . using and
, the steady state density matrix for intermediate temperatures can be written @xmath89 { } \nonumber \\ & \frac{n_0}{z_0 } \ , \mathfrak{t}(\gamma t \rightarrow \infty ) \int \limits_{-\infty}^{\infty } \int \limits_{-\infty}^{\infty } \exp \biggl\ { -\frac{m}{\hbar } \biggl [ \frac{1}{2 } \ , x_i^2 \
, \omega_0 \tanh \ ! \left(\frac{\beta_0 \hbar \omega_0}{2 } \right ) \nonumber \\ & \qquad \qquad \qquad \qquad \qquad + \frac{1}{2 } \ , y_i^2 \ , \omega_0 \coth \ !
\left(\frac{\beta_0 \hbar \omega_0}{2 } \right ) -i \xi x_i y_i - i
\xi x_f y_f \biggr ] \biggr\ } dx_i dy_i .\end{aligned}\ ] ] evaluating the gaussian integrals over @xmath33 and @xmath36 , then using the definitions of @xmath90 and @xmath88 simplifies this to @xmath91 { } 2 \mathfrak{t}(\gamma t \rightarrow \infty ) \sqrt{\frac{\pi \hbar}{m \omega_0 } \tanh \ ! \left(\frac{\beta_0 \hbar \omega_0}{2}\right ) } \frac{\omega_0}{\sqrt{\omega_0 ^ 2 + \xi^2 } } \nonumber \\ & \qquad \quad \cdot \exp \left(\frac{i}{\hbar } m \xi
x_f y_f \right ) .\end{aligned}\ ] ] it is instructive to use to rewrite in terms of the original wavefunction coordinates , @xmath14 and @xmath92 , @xmath93 { } 2 \mathfrak{t}(\gamma t \rightarrow \infty ) \sqrt{\frac{\pi \hbar}{m \omega_0 } \tanh \ ! \left(\frac{\beta_0 \hbar \omega_0}{2}\right ) } \frac{\omega_0}{\sqrt{\omega_0 ^ 2 + \xi^2 } } \nonumber \\ & \qquad \quad \cdot \exp \left[-\frac{m}{2 \hbar } \left(-i\xi q_f^2 + i \xi q_f^{\prime \ , 2 } \right ) \right ] \nonumber \\ & = 2 \mathfrak{t}(\gamma t \rightarrow \infty ) \sqrt{\frac{\pi \hbar}{m \omega_0 } \tanh \ ! \left(\frac{\beta_0 \hbar \omega_0}{2}\right ) } \frac{\omega_0}{\sqrt{\omega_0 ^ 2 + \xi^2 } } \sqrt{\frac{\pi \hbar}{m \xi } } \ , \varphi_0(q_f ) \varphi_0^\ast ( q_f ' ) , \end{aligned}\ ] ] where the wave function @xmath94 has been defined by _
0(q_f ) = ( ) ^ 1/4 ( q_f^2 ) .
therefore , in this higher temperature regime the equilibrium state of the harmonic oscillator is characterized by its ground state , but with the imaginary frequency @xmath95 .
if terms of order @xmath64 were not dropped , this frequency would be complex - valued in general . because the imaginary part of the frequency quantifies the decay rate into the stable ground state , all excited modes are unstable except for the ground state where the energy can always be chosen to be real @xcite . since
this decay rate is linearly related to the energy quantum number @xmath96 , the associated instability becomes more prominent with increasing excitation .
this result is interpreted as the reservoir acting to dissipate or extract energy from the oscillator , thereby preserving the coherence of a single quantum state .
through the decay of excited states , the environment can act to maintain the coherence of the oscillator at intermediate temperatures .
by contrast , at low or at very high temperatures the discriminant in becomes negative , thus changing the solutions of from exponential to oscillatory , y_cl(t ) = e^t \ { b ( t ) + c } , where the frequency is given by [ etadef ] = > 0 .
analysis that parallels that of the intermediate temperature case need not be repeated . instead
, corresponding expressions are obtained by the replacement @xmath97 in the previous treatment . using this substitution in and determines the analogous solution for this case , [ yclsolneta ] y_cl(t ) = \ { y_f e^(t-t ) ( t ) + y_i e^t } , with the corresponding end - point velocities given by @xmath98 \label{ydoteta0 } \\
\dot{y}_{cl}(t ) & \simeq \frac{\eta}{\sin ( \eta t ) } \left [ y_f\cos(\eta t ) - y_i e^{\gamma t } \right ] .
\label{ydotetat } \end{aligned}\ ] ] arriving at and follows from the fact that all calculations are performed in the severely underdamped regime for @xmath99 .
additionally , the expression for the integral of @xmath71 is obtained by first substituting into , then again using the transformation @xmath97 , @xmath100 \right\ } \nonumber \\ & \qquad \qquad \quad - \frac{1}{\omega^2 - ( \eta - i\gamma)^2 } \ , e^{(i\eta - \gamma)t } \
! \left\ { \cos[(\eta - i\gamma)t ] - \cos ( \omega t ) \right\ } \biggr ) \nonumber \\[.1 in ] & \qquad -2 y_f y_i \biggl ( \frac{1}{(\omega - \eta)^2 + \gamma^2 } \cos(\eta t ) \ ! \left\ { \cosh(\gamma t ) - \cos[(\omega - \eta)t ] \right\ } \nonumber \\ & \qquad \qquad \quad - \frac{1}{\omega^2 - ( \eta - i\gamma)^2 } \left\ { \cos[(\eta - i\gamma)t ] - \cos(\omega t ) \right\ } \biggr ) \nonumber \\[.1 in ] & \qquad \quad + y_i^2 \biggl ( \frac{1}{(\omega - \eta)^2 + \gamma^2 } \ , e^{\gamma t } \ !
\left\{\cosh(\gamma t ) - \cos [ ( \omega - \eta ) t ] \right\ } \nonumber \\ & \qquad \qquad \quad - \frac{1}{\omega^2 - ( \eta - i\gamma)^2 } \ , e^{(-i\eta + \gamma)t } \ !
\left\ { \cos[(\eta - i\gamma)t ] - \cos ( \omega t ) \right\ } \biggr ) \biggr]\nonumber \\ & + \ { \eta \rightarrow - \eta \ } .\end{aligned}\ ] ] due to the oscillatory nature of the solution for @xmath35 , the long - time approximations that led to do not apply for this case .
nonetheless , the damping parameter remains the smallest frequency in the system , thereby allowing terms of order @xmath64 and higher to be separated out as negligible .
hence , the dominant terms in the integrals over @xmath75 are easily obtained [ i1i2 ] @xmath101 \nonumber \\ & = \frac{\gamma}{\pi } re \left\ { \int \limits_{-2/\beta \hbar}^{2/\beta \hbar } \ ! d \omega \ , \omega \coth \ !
\left ( \frac{\beta \hbar \omega}{2 } \right ) \ !
\frac{1}{(\omega - \eta - i \gamma)(\omega - \eta + i \gamma ) } e^{i ( \omega - \eta ) t } \right\ } \nonumber \\ & = \eta \coth \ !
\left ( \frac{\beta \hbar \eta}{2 } \right ) \ !
e^{-\gamma t } + \ { \text{terms of order } \geq \gamma \ } , \label{i1 } \\[.1 in ] & \frac{\gamma}{\pi } \ !
\hbar}^{2/\beta \hbar } \ ! d \omega \ , \omega \coth \ !
\left ( \frac{\beta \hbar \omega}{2 } \right ) \ !
\frac{1}{\omega^2 - ( \eta - i\gamma)^2 } \cos(\omega t ) \nonumber \\ & = \ { \text{terms of order } \geq \gamma \ } .
\label{i2}\end{aligned}\ ] ] all remaining integrals appearing in are found from the @xmath102 forms of and .
using these results , the required integral involving @xmath71 becomes @xmath103.\end{aligned}\ ] ] substituting , and into the expression for the classical action leads to the result @xmath104 \biggr\ } .
\nonumber \\[-.08in]\end{aligned}\ ] ] as in the previous analysis , obtaining the effective action in terms of the end - point coordinates enables the evaluation of the integral @xmath105 appearing in the time - dependent density matrix . inserting and into the @xmath33 and @xmath106 integrals gives @xmath107 \biggr\ } dx_i dy_i .\end{aligned}\ ] ] performing the gaussian integrals along with the lengthy algebra involved leaves a simplified result @xmath108 \nonumber \\ & \qquad \qquad \qquad + 2i x_f y_f \biggl(\frac{\eta^2}{\omega_0 } - \omega_0 \biggr ) \eta \ , z_0 \cot(\eta t ) + x_f^2 \eta^2 \csc^2 ( \eta t ) e^{2 \gamma t } \biggr\ } \biggr ) , \end{aligned}\ ] ] in which the parameters @xmath109 and [ nudef ] = _ 0 z_0 + z_0 ^2(t ) + 2 z ^2 ( t ) ( t ) e^t , have been defined for brevity .
assuming @xmath64 to be sufficiently small , the renormalized frequency in can be approximated by [ eta = w0 ] _ 0 .
being in the highly underdamped regime , the expression for @xmath110 simplifies to [ nusimp ] _ 0
^2(_0 t ) , where we have inserted the explicit expressions for @xmath111 and @xmath112 .
likewise , also simplifies expression @xmath113 \biggr\ } .
\end{aligned}\ ] ] substituting this result into gives the time evolution of the full density matrix @xmath114 \biggr\ } .\end{aligned}\ ] ] since no long - time limits have been used in the derivation , equation includes the transient time behavior of the density matrix .
in addition to the expression for @xmath110 , this result shows the thermalization time to be on the order of the inverse damping parameter , @xmath115 , just like the previous case for intermediate temperatures . at times in excess of @xmath116 , @xmath110 simplifies to [ nulimit ] _ 0 ^2(_0 t ) ( ) e^2 t .
this limit demonstrates that when in contact with the thermalizing environment , the quantum oscillator equilibriates to a steady - state described by the density matrix @xmath117 \biggr\ } .
\nonumber \\[-.08in]\end{aligned}\ ] ] comparison with the initial density matrix of indicates that at low and at very high temperatures , for which the discriminant in is negative , the reservoir acts to thermalize the quantum oscillator to the usual boltzmann distribution of eigenstates .
the steady - state of a quantum oscillator can be dependent on the temperature of the environment to which it is coupled . modeled as an oscillator continuum
, the environment is quantified by a spectral density which is linear in frequency with a proportionality constant known as the coupling , or damping , parameter . because it is improbable to excite energies too far in excess of the mean thermal energy , the environmental spectrum has a cutoff of twice the mean thermal frequency , @xmath118 . in the path integral formulation ,
the steady - state of the oscillator is dependent upon the renormalized frequency , @xmath119 , which inherits its temperature dependence from the environmental cutoff . for low temperatures , the small frequency shift , @xmath120 , obtains the expected boltzmann thermalization .
to keep the renormalized frequency finite in the high - temperature limit , a damping parameter ( @xmath54 ) with a temperature cutoff , @xmath0 , is considered . if this cutoff is small enough such that the renormalized frequency is always real , then the expected boltzmann thermalization prevails at any temperature . however ,
if @xmath0 is sufficiently large , there is a range of temperatures for which the renormalized frequency becomes imaginary .
decaying at a rate given by the corresponding imaginary energy eigenvalue , all excited modes become unstable , whereas the ground state remains stable since its energy can always be chosen to be real .
dissipation now refers to an extraction of energy instead of a suppression of coherence .
concisely summarized by the example in fig.[dplot ] , the usual boltzmann thermalization prevails at low and at very high temperatures . for temperatures in between the environment
damps out the excited modes , thereby maintaining the coherence of a single quantum state . as indicated by experiments on photosynthetic molecules
@xcite and by recent experiments on superconducting graphite grains @xcite , there are systems where this intermediate damping regime could lie well within the room - temperature range . in particular , photosynthetic molecules display a quantum coherence having the same qualitative temperature dependence as described here .
such dependencies also arise in models of the exciton transport efficiency @xcite , thus suggesting a possible mechanism for a high - temperature cutoff in the damping parameter .
due to the appearance of hyperbolic functions , obtaining the dominant terms in the @xmath75 integrals of is somewhat more involved than in the very - high or in the lower temperature counterparts .
it is convenient to first consider the integral involving @xmath80 which contains two terms of the form @xmath121 \nonumber \\
\begin{split } & = \frac{1}{2 \pi i } \ !
\hbar}^{2/\beta \hbar } \ ! d \omega \ , \omega \coth \ !
\left ( \frac{\beta \hbar \omega}{2 } \right ) \ !
\left[\frac{1}{\omega - i(\xi + \gamma ) } - \frac{1}{\omega - i(\xi - \gamma ) } \right ] \\ & \qquad \qquad \qquad \qquad \qquad \qquad \cdot \left [ \cosh(\xi t ) \cos(\omega t ) + i \sinh(\xi t ) \sin(\omega t ) \right ] .
\end{split}\end{aligned}\ ] ] since this integral splits into two terms of opposite parity in @xmath64 , it suffices to evaluate only one , with the other obtained from the substitution @xmath122 .
a more compact expression can be written by denoting @xmath123 as the integral multiplying @xmath124 and @xmath125 as the integral multiplying @xmath126 , @xmath127 \nonumber \\ & \qquad \qquad = \cosh(\xi t ) ( i ) + i \sinh(\xi t ) ( ii ) .\end{aligned}\ ] ] reading off the integral identified as @xmath123 we have @xmath128 \ !
\left[\omega - i(\xi + \gamma ) \right ] } \cos(\omega t ) \nonumber \\ & = i(\xi + \gamma ) re \left\ { \int \limits_{-2/\beta \hbar}^{2/\beta \hbar } \ ! d \omega \ , \omega \coth \
! \left ( \frac{\beta \hbar \omega}{2 } \right ) \
! \frac{1}{\left[\omega + i(\xi + \gamma)\right ] \ !
\left[\omega - i(\xi + \gamma ) \right ] } \ , e^{i \omega t } \right\ } \nonumber \\ & = i(\xi + \gamma ) re \left\ { \vphantom{\int \limits_{{\cal c}_t } } 2 \pi i [ i
( \xi + \gamma ) ] \coth \
\frac{i \beta \hbar ( \xi + \gamma)}{2 } \right ] \ ! \frac{1}{2i ( \xi + \gamma ) } \ , e^{-(\xi + \gamma ) t } \right .
\nonumber \\ & \qquad \qquad \qquad \qquad \left .
- \int \limits_{{\cal c}_t } d \omega \ , \omega \coth \ !
\left ( \frac{\beta \hbar \omega}{2 } \right ) \
! \frac{1}{\omega^2 + ( \xi + \gamma)^2 } \ , e^{i \omega t } \right\ }
.\end{aligned}\ ] ] arriving at the last line requires an integration over contour @xmath129 , the positively - oriented semicircular arc of radius @xmath118 in the upper half complex plane . in the intermediate regime , temperatures are sufficiently high that [ a4 ] , ( ) ^-1 , thus allowing the integral over @xmath129 to be approximated as [ a5 ] _ c_t d ( ) e^i t _
c_t d ( ) e^i t .
the contour @xmath129 is more conveniently parametrized by the change of variables @xmath130 with these substitutions , the upper bound becomes @xmath131 where the small parameter @xmath132 has been defined as [ a8 ] = ( + ) 1 . to leading order , of the sum of the first integral and the last integral
is linear in @xmath132 , @xmath133 for the remaining term in , an upper bound can be established since [ a10 ] _ ^- d ( e^i ) e^i r e^i t < e^-2(+ ) t _ ^- d ( e^i ) e^i r ( ) t. terms of order @xmath134 are retained since these are used to obtain the corresponding terms of opposite parity in @xmath64 .
because all calculations are performed in the severely underdamped case , the damping parameter @xmath64 is by far the smallest frequency in the system .
consequently , the integral in may be neglected as it vanishes exponentially as @xmath135 in the required large - time limit . from these observations ,
the leading order contribution of the @xmath136 contour integral is [ a11 ] _
c_t d ( ) e^i t -2 ( 1 ) ( ) . using in , the dominant terms of integral @xmath123 are @xmath137 \ !
e^{-(\xi + \gamma)t } \\ & \quad + 2i \coth(1 ) \sin \ !
\left ( \frac{2t}{\beta \hbar } \right ) \ !
\beta \hbar ( \xi + \gamma ) + \mathcal{o } \ ! \left(\beta \hbar
\xi \right)^2 \biggr\}. \end{split}\end{aligned}\ ] ] the second integral of is identified as @xmath138 \ !
e^{-(\xi + \gamma)t } \nonumber \\ & \quad - 4 \coth(1 ) \cos \ ! \left(\frac{2t}{\beta \hbar } \right ) + \mathcal{o } \ ! \left(\beta \hbar
\xi \right ) \biggr\ } \ ] ] results and are used to express in terms of the required parameters @xmath139 \ !
e^{-\gamma t } + \mathcal{o } \ ! \left[\beta \hbar
\xi \sinh(\xi t)\right ] \biggr\ } \ ] ] as noted , the second contribution on the right - hand side of is derived by replacing @xmath64 with @xmath140 in .
subtracting these two contributions , then dividing by @xmath141 obtains @xmath142 \nonumber \\ & = \frac{i}{2 } ( \xi + \gamma ) \coth \ ! \left[\frac{i \beta \hbar ( \xi + \gamma)}{2 } \right ] \ !
e^{-\gamma t } - \frac{i}{2 } ( \xi - \gamma ) \coth \ ! \left[\frac{i
\beta \hbar ( \xi - \gamma)}{2 } \right ] \ !
e^{\gamma t } \nonumber \\[.1 in ] & \qquad -\frac{4 \gamma}{\pi } \ !
\left [ \coth(1 ) \cos \ ! \left(\frac{2t}{\beta \hbar } \right ) + \mathcal{o } \ !
\left ( \beta \hbar \xi \right ) \right ] \ !
\sinh(\xi t ) .\end{aligned}\ ] ] equation along with its @xmath102 form are used to obtain the @xmath80 integral in , @xmath143 \ !
+ ( \xi - \gamma ) \coth \ ! \left[\frac{i
\beta \hbar ( \xi - \gamma)}{2 } \right ] \biggr\ } \sinh(\gamma t ) \nonumber \\ & \qquad \qquad + \frac{4 \gamma}{\pi } \left [ \coth(1 ) \cos\left(\frac{2t}{\beta \hbar } \right ) + \mathcal{o } \ !
\left ( \beta \hbar \xi \right ) \right ] \sinh(\xi t ) .\end{aligned}\ ] ] also in is the integral of @xmath81 which is a linear combination of two terms of the form @xmath144 \ ! \left[\omega - i(\xi - \gamma ) \right ] } \ , e^{i \omega t } \right\ } \nonumber \\ & = \frac{\gamma}{\pi } re \left\ { \vphantom{\int \limits_{{\cal c}_t } } \pi i \coth \ ! \left
[ \frac{i \beta \hbar ( \xi - \gamma)}{2 } \right ] \ !
e^{-(\xi - \gamma ) t } \right .
\nonumber \\ & \qquad \qquad \qquad \qquad \qquad \left .
- \int \limits_{{\cal c}_t } d \omega \ , \omega \coth \ !
\left ( \frac{\beta \hbar \omega}{2 } \right ) \
! \frac{1}{\omega^2 + ( \xi - \gamma)^2 } \ , e^{i \omega t } \right\ } \nonumber \\ & = \frac{\gamma}{\pi } re \biggl\ { \pi i \coth \ ! \left [ \frac{i \beta \hbar ( \xi - \gamma)}{2 } \right ] \ !
e^{-(\xi - \gamma ) t } + 2 \coth(1 ) \sin \ ! \left(\frac{2t}{\beta \hbar } \right ) \ !
\beta \hbar ( \xi - \gamma ) \nonumber \\ & \qquad \qquad \qquad \qquad \qquad \qquad \vphantom{\left(\frac{2t}{\beta \hbar } \right ) } + \mathcal{o } \ !
\xi \right)^2 \biggr\ } \nonumber \\[.1 in ] \begin{split } & = i \gamma \coth \
\frac{i \beta \hbar ( \xi - \gamma)}{2 } \right ] \ ! e^{-(\xi - \gamma)t } + \frac{2 \gamma}{\pi } \coth(1 ) \sin \ !
\left(\frac{2t}{\beta \hbar } \right ) \ !
\beta \hbar(\xi - \gamma ) \\ &
\qquad \qquad \qquad \qquad \qquad \qquad + \gamma \ , \mathcal{o } \ ! \left ( \beta \hbar \xi \right)^2 .
\end{split}\end{aligned}\ ] ] arriving at the third equality requires the expression for the integral over @xmath129 given by .
similar to the previous case , equation and its @xmath102 form are used to obtain the @xmath81 integral in , @xmath145 \sinh\left [ ( \xi - \gamma ) t \right ] - \frac{2 \gamma}{\pi } \coth(1 ) \sin \ ! \left ( \frac{2t}{\beta
\hbar } \right ) \ ! \beta \hbar
( \xi - \gamma ) \nonumber \\ & \qquad \qquad + \gamma \sinh^2 \left [ \frac{1}{2 } ( \xi - \gamma ) t \right ] \mathcal{o}(\beta \hbar \xi)^2 .\end{aligned}\ ] ] equations and express the integrals over @xmath75 in terms of the required parameters .
inserting these results back into gives @xmath146 \nonumber \\ & \qquad \qquad \qquad \ , \ , + y_i^2 \ , e^{\gamma t } \left ( \zeta - \varepsilon_+ e^{-\xi t } - \varepsilon_-e^{\xi t } \right ) \bigr\ } , \end{aligned}\ ] ] where @xmath147 \ !
\sinh ( \gamma t ) + i(\xi - \gamma ) \coth \
\frac{i \beta \hbar ( \xi - \gamma)}{2 } \right ] \ ! \sinh ( \gamma t ) \nonumber \\ & \quad + \frac{8 \gamma}{\pi } \sinh(\xi t ) \mathcal{o } ( \beta \hbar \xi ) \\[.15 in ] \varepsilon_\pm & = i \gamma \coth \ ! \left [
\frac{i \beta \hbar ( \xi \pm \gamma)}{2 } \right ] \ ! \sinh
\left[(\xi \pm \gamma ) t)\right ] + \frac{2 \gamma}{\pi } \coth ( 1 ) \sin \ !
\left(\frac{2t}{\beta \hbar } \right ) \beta \hbar ( \gamma \pm \xi ) \nonumber \\ & \quad + \gamma \sinh^2\left [ \frac{1}{2 } ( \xi \pm \gamma ) t \right ] \mathcal{o } ( \beta \hbar \xi)^2 .
\end{aligned}\ ] ] the terms of opposite parity in @xmath78 , denoted by @xmath148 , have been explicitly included in @xmath149 and in @xmath150 . for times larger than or equal to the thermalization time ,
the limit @xmath83 is applicable for which the dominant terms in @xmath150 and @xmath149 are [ a21 ] @xmath151 { } \frac{4 \gamma}{\pi } \ , e^{\xi t } \ , \mathcal{o }
( \beta \hbar \xi ) \\
\varepsilon_\pm & \xrightarrow[\gamma t \rightarrow \infty ] { } \frac{\gamma}{2 } \ , e^{(\xi \pm \gamma ) t } \biggl\ { i \coth \
\frac{i \beta \hbar ( \xi \pm \gamma)}{2 } \right ] + \mathcal{o } ( \beta \hbar \xi)^2 \biggr\ } . \end{aligned}\ ] ] finally , expressions are used to obtain the thermalization time limit of [ arintsimpxia4 ] _ 0^t
y_cl(t ) _ 0^t
a_r(t-t ) y_cl(t ) dt
dt i ( ) ( y_f^2 + y_i^2 ) .
in addition to dropping terms of order @xmath152 and higher , it has been assumed that @xmath153 , consistent with the severely underdamped case .
+ this result is similar to that obtained for an isolated quantum oscillator where complex coordinates along with a complex - valued hamiltonian are introduced at the outset .
see s. g. rajeev , ann .
( 2007 ) 1541 - 1555 .
|
the feynman - vernon path integral formalism is used to derive the density matrix of a quantum oscillator that is linearly coupled to an environmental reservoir .
although low - temperature reservoirs thermalize the oscillator to the usual boltzmann distribution , reservoirs at intermediate temperatures reduce this distribution to a single , coherent ground state .
associated with this state is an imaginary frequency indicating an environment which absorbs energy from the oscillator through the suppression of all excited modes .
further increase of the environmental temperature results again in the thermalization of the quantum oscillator to the expected boltzmann distribution .
qualitatively , this result could account for high - temperature quantum effects including the superconducting properties of graphite grains as well as the quantum coherence observed in photosynthetic systems .
decoherence , thermalization , imaginary eigenvalue , excited state decay , photosynthesis , high - temperature superconductivity
|
the acceleration of particles at a decelerating relativistic collisionless shock front constitutes a key building block of the afterglow model of gamma - ray bursts ( grb , mszros & rees 1997 ) .
the standard phenomenology models the accelerated electron population as powerlaw @xmath1 , which radiates powerlaw photon spectra of the form @xmath2 , with a temporal decay index @xmath3 and a frequency index @xmath4 that are direct functions of @xmath5 , see e.g. piran ( 2005 ) for a review , or e.g. sari et al .
( 1998 ) , panaitescu & kumar ( 2000 ) for detailed formulae . from both microscopic and observational points of view
, the situation however appears more complex , in spite of several remarkable results of the past decade . on the microscopic level , for instance ,
one understands the formation of a relativistic collisionless shock front in a weakly magnetized medium such as the interstellar medium ( ism ) through the self - generation of intense small scale electromagnetic fields that act as the mediating agents for the transition from the far upstream unshocked state to the far downstream shocked state . the accelerated particles , as forerunners of the shock front , play a central role in triggering the microinstabilities that build the self - generated field .
in turn , this self - generated microturbulence controls the scattering of these particles and it therefore directs the acceleration process , which becomes intricately non - linear .
this general scheme has been validated so far in high performance particle - in - cell ( pic ) simulations ( e.g. spitkovsky 2008 , keshet et al .
2009 , sironi & spitkovsky 2009 , 2011 , martins et al . 2009 ) and understood on the basis of analytical arguments at the linear level ( e.g. medvedev & loeb 1999 , gruzinov & waxman 1999 , lyubarsky & eichler 2006 , lemoine & pelletier 2010 , 2011a , rabinak et al .
the situation becomes more complex when one tries to bridge the gap between the limited simulation timescales and the much longer timescales probed by the observations .
on such timescales , one would indeed expect that the microturbulence has died away ( e.g. gruzinov & waxman 1999 ) , yet grb afterglow modelling has generally pointed to a field strength close to a percent of equipartition permeating the blast , on day timescales .
the origin of this field and its relation with the microturbulence behind the shock front has remained a nagging issue for many years . on the observational level ,
the recent era of rapid follow - up observations in the x - ray and gev domain has brought its wealth of surprises .
the swift satellite has revealed x - ray afterglow light curves that differ appreciably in the @xmath6s domain from the canonical afterglow model ( nousek et al .
2006 , obrien et al .
2006 ) . of more direct interest
to the present work , the _ fermi_-lat telescope has reported the discovery of long - lived ( @xmath7s ) gev emission in a fraction of observed bursts . in one case ( grb090510 ) , this emission has been measured almost contemporarily to emission in the x - ray and optical domains as early as @xmath8s ( ackermann et al .
2010 , de pasquale et al .
this long - lived high energy emission has been shown to fit nicely the predictions of a model in which the electrons cool slowly through synchrotron radiation in a background shock compressed magnetic field , without any need for microturbulence ( barniol - duran & kumar 2009 , 2010 , 2011a ; see also gao et al .
2009 ; de pasquale et al . 2010 ; corsi et al .
2010 ; ghirlanda et al . , 2010 ;
he et al 2011 ; and see ghisellini et al .
2010 , razzaque 2010 and panaitescu 2011 for alternative points of view ) . given the past history in grb afterglow modelling , such a low magnetization of the blast may come as a surprise , but it may also point out that after all , the microturbulence does decay away as theoretically expected , and that the high level of turbulence seen on day timescales has been seeded through some other instability , but this comes at the price of an extraordinarily low external density @xmath9@xmath10 .
this interpretation is not considered here , see also sec .
[ sec : fcons ] for further discussion . ] .
depending on how fast and how far from the shock this microturbulent layer decays , it is likely to influence the particle energy gains from fermi acceleration , and losses through synchrotron radiation .
the microturbulent layer must actually ensure the scattering of accelerated particles , because in the absence of microturbulence , these particles would be advected away with the transverse magnetic field lines to which they are tied and acceleration would not take place ( e.g. begelman & kirk 1990 , lemoine et al . 2006 , niemiec et al .
2006 , pelletier et al .
now , scattering in small scale turbulence is so slow that producing gev photons at an external blast wave of lorentz factor of a few hundreds represents a challenge , see e.g. kirk & reville ( 2010 ) , lemoine & pelletier ( 2011c ) , and see also piran & nakar ( 2010 ) , sagi & nakar ( 2012 )
. it would be about impossible if the particles were to scatter in the microturbulent layer then radiate in the much weaker shock compressed background field .
therefore , the interpretation of barniol - duran & kumar ( 2009 , 2010 , 2011a ) actually suggests that the microturbulence also plays a role in the radiation of gev photons , at the very least , if not in shaping the synchrotron spectra over the broad spectral range . in other words , the observation of extended gev emission and its early follow - up in other wavebands may be opening a rare window on the dynamics of the microturbulence in weakly magnetized relativistic collisionless shocks . in this context , the present paper proposes to discuss the synchrotron spectra and more generally the afterglow spectrum @xmath11 of a decelerating relativistic blast wave , accounting for the time evolution of the microturbulence behind the shock front . while the initial motivation of this work was to provide a concrete basis for the scenario of barniol - duran & kumar ( 2009 , 2010 , 2011a ) , in which particles scatter in a time decaying microturbulence but radiate in a region devoid of microturbulence , it has become apparent that the possibility of decaying microturbulence opens a quite rich diversity of phenomena , which deserve to be discussed in detail
this problem has been tackled by rossi & rees ( 2003 ) , who considered the simplified case of a homogeneous microturbulent layer that dies instantaneously beyond some distance , and by derishev ( 2007 ) , who showed that a particle radiating in a time evolving magnetic field can lead to spectra quite different from the standard one - particle synchrotron spectra .
borrowing from the latest pic simulations , the present paper establishes a model of the microturbulence strength and coherence length evolving as powerlaws in time beyond some distance , until the decay saturates down to the background shock compressed magnetic field ; the present study then calculates the synchrotron spectra in this structure for various typical configurations ( slow cooling , fast cooling , with and without inverse compton losses ) and it discusses the problem of particle scattering , acceleration timescale and maximum photon energy in this setting . as such , it generalizes and encompasses these former studies , in the spirit of providing new tools with which one can analyse existing and forthcoming data . a brief comparison to present early afterglow observations
is provided .
the detailed spectra and the spectro - temporal indices @xmath3 , @xmath4 of @xmath12 are provided in appendix [ sec : appfnu ] , while section [ sec : synchspec ] details the model for the evolution of microturbulence and provides the general characteristics of the afterglow light curves and spectral energy distributions in various configurations .
section [ sec : disc ] discusses the scattering process and the maximal acceleration energy , and it confronts the above models to existing data .
the results are summarized in sec .
[ sec : concl ] . throughout , this paper adopts the standard notation @xmath13 with @xmath14 a generic quantity in cgs units .
fiducial values used for numerical applications correspond to those derived in barniol - duran & kumar ( 2009 , 2010 , 2011a ) , and he et al .
( 2011 ) , e.g. an external density @xmath15@xmath10 , a blast lorentz factor @xmath16 at @xmath8s and it is assumed that the blast has entered the deceleration regime .
one must distinguish the time @xmath17 in the observer frame , also written @xmath18s in numerical applications , from the time experienced by a particle since shock entry ; this difference is manifest everywhere . for convenience ,
the notation @xmath19 is introduced , @xmath20 denoting the redshift of the grb .
the self - generation of microturbulence in the precursor of a relativistic collisionless shock front propagating in a very weakly magnetized medium appears both guaranteed and necessary to the maintenance of the shock .
it is necessary because in the absence of a background magnetic field , self - magnetization is required to build up a magnetic barrier that initiates the shock transition .
it is guaranteed because the development of microinstabilities follows naturally from the penetration of the unshocked plasma by the anisotropic beam of supra - thermal particles moving ahead of the shock ( e.g. medvedev & loeb 1999 ) .
studies of non - relativistic collisionless magnetospheric shocks have shown that dissipation ahead of the shock is initiated by the reflection of ambient particles ( i.e. from the unshocked plasma ) on the shock front in the compressed magnetic field ( e.g. leroy et al .
particle - in - cell simulations indicate that a similar phenomenon takes place at a relativistic unmagnetized shock , although the particles now reflect on the small scale electromagnetic fields self - generated by the microinstabilities ( e.g. spitkovsky 2008 ) .
the reflected and accelerated particle populations merge together and trigger microinstabilities such as the weibel ( filamentation ) instability or oblique electrostatic instabilities , provided the precursor extends far enough for these modes to grow on the precursor crossing timescale ( lemoine & pelletier 2010 , 2011a ) . in a very weakly magnetized shock wave , with magnetization typical of the ism and blast lorentz factor @xmath21 , this appears guaranteed .
as seen from the shock frame ( in which the shock front lies at rest ) the incoming kinetic energy is carried by the protons , the electrons carrying only a fraction @xmath22 of the incoming flow kinetic energy .
energy transfer between the two species in the microturbulence leads to heating of the electron population , close to equipartition by the time it reaches the shock front , as observed in current pic simulations ( sironi & spitkovsky 2011 ) , see also the discussion in lemoine & pelletier ( 2011a ) .
equipartition means that the incoming electrons carry lorentz factor @xmath23 , hence their skin depth scale ( downstream frame ) @xmath24^{1/2}\sim c/\omega_{\rm pi}$ ] .
the natural length scale of the electromagnetic structures produced by these microinstabilities is therefore the ion skin depth scale @xmath25 of the upstream plasma .
particle - in - cell simulations not only validate the above general scheme , they also provide interesting constraints on the shape and evolution of microturbulence ahead and behind the shock front .
two most recent and most detailed studies are of direct interest to the present work .
chang et al . (
2008 ) have performed long simulations of the evolution of micro - turbulence behind a relativistic shock front .
the simulations have been computed for an unmagnetized pair plasma with relative lorentz factor @xmath26 between upstream and downstream . in weakly magnetized relativistic shock waves
, the preheating of electrons in an electron - ion shock of similar configuration appears so efficient that for all practical matters , the downstream plasma behaves as a relativistic pair plasma ; hence the results of chang et al . (
2008 ) can be transposed to an electron - ion shock .
these simulations show that the microturbulence remains mostly static in the downstream rest frame , and that it is composed of an intermittent magnetic field structure that can be roughly described as a collection of magnetic loops and islands on typical length scales @xmath27 .
one clear observation made in this work is that the small scale structures dissipate first , leaving the large scale clumps unaffected over the timescale of the simulation .
chang et al . ( 2008 ) interpret this gradual erosion as collisionless damping , with a damping frequency @xmath28 ( @xmath29 denoting the spatial scale ) .
if the magnetic turbulence is described in fourier space as a power law spectrum with most of magnetic power on small length scales , this implies a decay of the magnetic field strength accompanied by an evolution of the coherence scale @xmath30 .
this is made explicit further below .
the longest pic simulation so far for a relativistic shock is that of keshet et al .
( 2009 ) , which extends to about @xmath31s ( comoving time ) . for values envisaged by barniol - duran & kumar ( 2009 , 2010 , 2011a ) and he et al .
( 2011 ) to describe the gev extended emission , i.e. an ejecta of energy @xmath32erg , launched into a medium of density @xmath15@xmath10 with initial lorentz factor @xmath33 , the above timescale represents close to @xmath34% of a dynamical timescale at an observer time of a hundred seconds .
keshet et al .
( 2009 ) provide a detailed study of the magnetic field power spectrum of the turbulence and its evolution .
they confirm most of the findings of chang et al .
( 2008 ) ; in particular , they show that the magnetic field does decay behind the shock wave , but on rather long length scales compared to a skin depth @xmath25 .
more importantly , they find that the presence of shock accelerated particles influences the decay timescale of the magnetic field with a general trend being that higher energy particles ensure a longer lifetime for the downstream microturbulence .
given that the simulations of keshet et al .
( 2009 ) extend for a time that is much smaller than the dynamical time , it has not had time to produce very high energy particles and to probe their impact .
such high energy particles would tend to populate the magnetic perturbation spectrum with longer wavelength modes , which would then decay on longer timescales when downstream . in any event , this should not call into question what has been said above , since low energy particles carry most of the energy of a shock accelerated population with index @xmath35 .
nevertheless , to probe how far the perturbation spectrum may be populated , one can conduct the following exercise .
the maximal size of the precursor is given in a reasonable approximation by @xmath36 , where @xmath37 represents the gyration radius of the highest energy ions in the background upstream magnetic field , and @xmath38 the shock lorentz factor as measured upstream .
this result is discussed in detail in plotnikov et al .
( 2012 ) but it can be understood as follows .
the highest energy ions are those that travel the furthest away from the shock , since electrons are generally accelerated to a smaller energy due to synchrotron losses ; furthermore , the particles gyrate by an angle @xmath39 over a timescale @xmath40 before being caught up by the shock front ( achterberg et al .
2001 ) ; finally , the typical distance between the shock front and the particle is @xmath41 . assuming that the ions are accelerated on a timescale @xmath42 ( see also sec .
[ sec : accel ] ) and comparing the time available for acceleration with the age of the shock wave @xmath43 , one finds the precursor size @xmath44 .
this indicates that the perturbation may well extend on several decades . in the absence of ions ,
the precursor size would be set by the highest energy electrons ; balancing acceleration at a bohm rate ( as experienced downstream ) to synchrotron losses , one would find a precursor size about 20 times smaller for the adopted fiducial parameters , nevertheless much larger than the typical size of the fluctuations . finally , keshet et al .
( 2009 ) suggest a damping frequency @xmath45 , which implies , when combined with the above result that the decay of the microturbulence might extend over quite long spatial scales . following chang et al .
( 2008 ) , the magnetic field power spectrum is described as a time decaying powerlaw form in the downstream ( comoving ) frame , with @xmath46 with @xmath47 denoting the time ( downstream frame ) since shock entry of the corresponding plasma element , @xmath48 ( resp .
@xmath49 ) the minimum ( resp .
maximum ) wavelength scale of the microturbulence at @xmath50 , @xmath51 so that the turbulent power lies at the smallest scales , @xmath52 for normalization purposes , @xmath53 denotes the rms field strength at @xmath50 and @xmath54 the damping time , which depends on @xmath29 : @xmath55 assuming @xmath56 for the moment , eq .
( [ eq : bturb ] ) can be integrated in terms of an incomplete gamma function , @xmath57\right\}\ , \nonumber\\ & & \label{eq : dbt}\end{aligned}\ ] ] with @xmath58 for convenience , one may approximate eq .
( [ eq : dbt ] ) with respectively the small and large argument limits to describe the evolution of the magnetic field strength as @xmath59 with the following definition @xmath60 in the following , the numerical factor @xmath61 , of order unity , will be dropped henceforth .
equation [ eq : alphat ] shows that the temporal decay index of the magnetic field behind the shock is inherently linked to how power is distributed on scales larger than the minimum scale @xmath62 as characterized by @xmath63 and to how fast small scale features are dissipated as characterized by @xmath64 .
the interpretation for this is clear : as small scales are erased , magnetic power is removed , but the rate at which the total strength erodes depends on how much strength is left at longer wavelengths . in fine , the uncertainty on @xmath65 is related to the sourcing of large wavelength fluctuations , which are likely related to the dynamics of high energy particles in the upstream .
the above assumption @xmath56 has been discussed above .
its robustness depends crucially on the influence of maximal energy particles upstream of the shock front . in the extreme opposite case @xmath66
, one should observe a roughly constant magnetic field while @xmath67 , followed by fast decay once @xmath68 .
this situation may be accounted for by eq .
[ eq : bevol ] with a more pronounced value of @xmath65 .
the following therefore considers a range of possibilities for @xmath65 , even though the pic simulations of chang et al .
( 2008 ) and keshet et al .
( 2009 ) both suggest @xmath69 . more specifically ,
chang et al . (
2008 ) suggest that the magnetic field fourier spectrum ( for @xmath70 , not @xmath71 ) in wavenumber has slope @xmath72 , which corresponds to @xmath73 , and @xmath74 leading to @xmath75 .
keshet et al .
( 2009 ) show that right behind the shock front , the magnetic field decays exponentially on short a distance scale to level off at a strength corresponding to @xmath76 for some hundreds of skin depth as a fiducial value , even though the magnetic energy density reaches @xmath77% of the incoming energy at the shock transition itself ( see chang et al . 2008 , keshet et al . 2009 ) . ] .
a closer inspection of their fig.3 however reveals that the initial exponential decay leaves way to a powerlaw decay at late simulation times ( thus meaning far downstream ) and by eye , one estimates @xmath78 .
these simulations thus indicate a value of @xmath65 between @xmath79 and @xmath80 ; however , given the present limitations of the pic simulations , and the above possible caveat related to the extension of the magnetic perturbation spectrum , one can not exclude yet that @xmath81 . in this respect
, one must point out that recent simulations of the development and the dynamics of relativistic weibel turbulence indeed suggest a value @xmath82 ( medvedev et al .
although these simulations do not simulate the shock itself , but a weibel turbulence through the interpenetration of two relativistic beams , these are 3d while the shock simulations of chang et al .
( 2008 ) and keshet et al .
( 2009 ) are 2d . in order to account for these different possibilities in the following , eq .
[ eq : bevol ] is kept in its present form , but the decay exponent @xmath65 is assumed to take possibly mild or more pronounced values . depending on whether @xmath81 or @xmath69 , it will be seen that radically different radiative signatures are to be expected . in summary , theoretical arguments combined with recent high performance pic simulations
suggest the following characterization for the evolution of the microturbulence behind a relativistic ( weakly magnetized ) shock front .
immediately behind the shock , the magnetic field carries strength @xmath53 corresponding to an equipartition parameter @xmath83 with fiducial value @xmath84 , while @xmath85 represents the fiducial value for the coherence scale at that same location .
the magnetic field strength decays as @xmath86 after a time @xmath87 defined through @xmath88 , of the order of hundreds to thousands of inverse plasma times , until it eventually settles at the shock compressed value @xmath89 . in the following ,
this timescale is rewritten in units of inverse plasma times as @xmath90 meaning also that the undecayed part of the microturbulence extends for @xmath91 skin depths .
note that @xmath92 according to the above simulations .
finally , the coherence length of the microturbulence evolves as @xmath93 , with fiducial value @xmath94 .
as a particle gets fermi accelerated , it interacts with the turbulent layer within a scattering length scale @xmath95 of the shock front .
this scattering length scale controls the residence time hence the acceleration timescale hence the maximal energy that can be reached ; it is discussed in more detail in sec .
[ sec : accel ] . for the time
being , it suffices to note that the larmor radius of the bulk of the electrons , with minimum lorentz factor @xmath96 is so small that these electrons can only explore the undecayed part of the turbulence : @xmath97 the microturbulence thus controls the acceleration of the bulk of electrons , independently of how fast this microturbulence decays or how large the blast lorentz factor may be . at lorentz factors @xmath98 of interest for high energy radiation , the electrons may start to explore the region @xmath68 .
then the transport becomes non trivial ; its impact on acceleration is discussed in sec .
[ sec : accel ] . during the acceleration stage ,
the particle moves in a near ballistic manner and diffusive effects can be neglected , given that the return probability decreases fast with the number of steps of length @xmath99 taken .
therefore , as a particle moves away from the shock on a distance scale @xmath100 along the shock normal during a time @xmath101 , with @xmath102 the angle to the shock normal , it explores a microturbulence that has decayed according to the laws given above with @xmath103 .
this factor of @xmath104 of course results from the convective velocity @xmath105 of the downstream plasma .
on length scales much larger than @xmath95 , a particle diffuses in the microturbulence and in a first approximation , one can describe its transport by advection with the downstream plasma .
for such particles , @xmath106 .
the above slight difference between @xmath47 and @xmath107 does not impact the results given further below and can be neglected in view of the uncertainties related to the time evolution of the turbulence . in the following , @xmath47 and @xmath107
are thus be used interchangeably .
the cooling history of an electron of lorentz factor @xmath108 obeys the standard law @xmath109 with @xmath110 the compton cooling factor . for the time being , one considers the simple case @xmath111 ; the influence of inverse compton losses is discussed in detail in app .
[ sec : appfnu ] and further below .
one then defines a lorentz factor @xmath112 such that , if @xmath113 , the particle cools in the undecayed microturbulence where @xmath67 ( at time @xmath114 ) , while if @xmath115 , the particle does not cool in that layer but further on . writing the synchrotron cooling time for particles of lorentz factor @xmath108 in a ( constant ) magnetic field of strength @xmath70 as @xmath116 $ ] , the lorentz factor @xmath112 can also be defined as the solution of @xmath117\equiv 1\ .\label{eq : gmu+}\ ] ] in terms of the fiducial values of interest here , @xmath118 with @xmath119 .
this value of @xmath112 generally exceeds the maximal lorentz factors that can be achieved through shock acceleration , therefore particles cool outside this undecayed turbulent layer , unless @xmath120 is larger than expected , as discussed in sec .
[ sec : accel ] .
the latter may well happen , if for instance @xmath121 and/or @xmath122 . if a particle exits the acceleration process with a lorentz factor @xmath123 , then at time @xmath47 , @xmath124\gamma_{e,0}/\gamma_{\mu+ } } } & ( \mu(t)\gg1)\ .
\end{cases } \label{eq : cool2}\ ] ] if @xmath125 , the particle cools down to @xmath112 within the layer where @xmath67 ( i.e. @xmath114 ) and subsequently , it continues cooling if @xmath69 or stops its cooling if @xmath81 . if @xmath126 on the contrary , the particle either cools later in the decaying microturbulence if @xmath69 , or not if @xmath81
. in any case , the particle of course eventually cools in the background shock compressed field ( notwithstanding issues related to the available hydrodynamical timescale ) . whichever occurs influences the afterglow light curve and spectral energy distribution .
if @xmath69 , it is convenient to define a second lorentz factor , @xmath127 , as the lorentz factor for which cooling occurs on a timescale @xmath128 such that @xmath129 , i.e. at the time at which the turbulence field has relaxed to the background shock compressed value @xmath130 . for the time being , no consideration is made of the hydrodynamical timescale of the blast . then , if @xmath131 ( and @xmath132 ) , most particles cool in the decaying microturbulent layer .
one finds @xmath133 in terms of the upstream magnetization parameter @xmath134 .
the lorentz factor @xmath127 depends exponentially on @xmath65 ; it may therefore take very different values . to calculate the radiative signature ,
one integrates over the cooling history of the electron population , as in gruzinov & waxman ( 1999 ) , although for simplicity , the calculation is done in a one - dimensional quasi - steady state approximation , meaning that the secular hydrodynamical evolution of the blast is neglected in the course of this integration over the blast width .
this method corresponds to the steady state approximation of sari et al .
( 1998 ) , when calculating the stationary electron distribution in a homogeneous shell .
the spectral power density of the blast can then be written in the downstream frame in terms of an integral over a particle history , up to a ( comoving ) hydrodynamical timescale @xmath135 ( see eq .
[ eq : tdyn ] ) @xmath136 with @xmath137 the spectral power density radiated by an electron at time @xmath47 , of initial lorentz factor @xmath123 and of cooling history given by eq .
( [ eq : cool2 ] ) ; see also eq .
[ eq : appdfnu ] .
the above expression is folded over the injection distribution @xmath138 , which is assumed to take a power law form between @xmath139 and @xmath140 : @xmath141 with @xmath142 the number of electrons swept and shock accelerated by the shock wave per unit time , as measured in the downstream frame .
the detailed calculation of the spectral power @xmath143 is carried out in appendix [ sec : appfnu ] for different relevant cases . in particular ,
two distinctions have to be made : whether the microturbulence decays rapidly beyond @xmath144 or not and whether inverse compton losses contribute significantly to the cooling of electrons .
these cases are examined in turn in the next subsections , in parallel to the discussion of app .
[ sec : appfnu ] . as discussed in app .
[ sec : appfnu ] , the gradual evolution of the microturbulence behind the shock affects the spectro - temporal flux @xmath145 in various ways . for one ,
particles of different lorentz factors cool in different magnetic fields , with particles of lower energy experiencing lower magnetic fields at cooling .
this implies that the characteristic synchrotron frequencies are modified and more specifically , that ratios of characteristic frequencies are stretched with respect to the standard case of a homogeneous turbulent layer .
regarding the cooling frequency @xmath146 , the modification is non - trivial because the cooling lorentz factor @xmath147 itself depends in a non - trivial way on the temporal decay index of the turbulence , see eq .
[ eq : gammac2 ] . the characteristic frequency @xmath148 associated to particles of lorentz @xmath139
can also be modified in a non - trivial way , since @xmath149 $ ] , with @xmath150 defined as the synchrotron peak frequency of particles of lorentz factor @xmath139 in a magnetic field of strength @xmath151 , see eq .
[ eq : nusyn ] .
the magnetic field @xmath151 in which the particles of lorentz factor @xmath139 radiate most of their energy takes the shock compressed value @xmath152 if both @xmath153 and @xmath154 , but @xmath155 if @xmath156 and @xmath157 , or @xmath158 , with @xmath159 $ ] defined in eq . [
eq : delta ] .
the ratio @xmath160 determines whether the turbulence has relaxed to @xmath152 by the back of the blast or not , and this value depends exponentially on @xmath65 : @xmath161 in practice , it can take small or large values at different times , even for @xmath162 for which the numerical prefactor becomes @xmath163 . in direct consequence of the above , the deceleration of the blast implies a non - trivial temporal evolution of the characteristic frequencies .
this modifies the standard temporal evolution of @xmath145 .
furthermore , as a particle gets advected away from the shock with the microturbulence , it radiates at decreasing frequencies , whether it cools efficiently or not .
consequently , the changing magnetic field also modifies the spectral slope of the flux @xmath145 .
appendix [ sec : appfnu ] provides a detailed discussion of the possible cases , with detailed expressions for the characteristic frequencies @xmath148 , @xmath146 , @xmath164 and @xmath165 , depending on their respective orderings . following app .
[ sec : appfnu ] , the present discussion does not consider the cases in which either @xmath166 or @xmath167 , because such cases appear rather extreme in terms of the parameters characterizing the turbulence , as discussed in the former section .
moreover , these cases tend to the standard model of a synchrotron afterglow in a homogeneous turbulence when both @xmath168 and @xmath169 and they can be easily recovered from app .
[ sec : appfnu ] .
( top red curve ) , and in a decaying microturbulence such that @xmath162 , @xmath170 . for both ,
the blast parameters are @xmath171 , @xmath172 and the injection distribution index @xmath173 .
the regime is slow cooling , and @xmath156 ( corresponding to case 1 of fig .
[ fig : spec_asupm1 ] for the microturbulent model ) .
[ fig : spec1 ] , scaledwidth=49.0% ] figure [ fig : spec1 ] presents a concrete example of a spectral energy distribution , comparing the standard prediction for a homogeneous turbulence with @xmath174 ( blue line ) to a time evolving microturbulence with @xmath162 , also starting at @xmath174 ( red line ) , with @xmath170 .
both models assume @xmath175 corresponding to an observer time @xmath176s for a blast at @xmath177 with @xmath178 , @xmath172 , an injection slope @xmath173 and a circumburst medium of constant density .
inverse compton losses are neglected throughout the blast in this example .
figure [ fig : spec1 ] reveals the characteristic stretch of frequency range , with @xmath179hz for the dynamical microturbulent model ( resp .
@xmath180hz in the homogeneous turbulence ) and @xmath181hz ( resp .
@xmath182hz ) .
the frequency @xmath183hz lies outside the range of fig .
[ fig : spec1 ] . for the above case of slow cooling , the spectral indices at low and intermediate frequencies , respectively
@xmath184 and @xmath185 ( with @xmath186 ) remain unaffected in the presence of decaying microturbulence .
however the temporal index @xmath3 is modified in these cases , even at low frequencies , which opens the possibility of testing such cases through the temporal behavior of an early follow - up in the optical .
section [ sec : grb ] below offers a comparison of such spectra with the observed light curve of grb090510 .
regarding the fast cooling part of the electron population , both the spectral and temporal indices are modified , see table [ tab : gradnoic ] for their detailed values .
see also fig .
[ fig : spec_asupm1 ] for an illustration of possible synchrotron spectra .
there are two main differences between the synchrotron spectra of gradually vs rapidly decaying microturbulence . in the former case
, particles may cool in the decaying microturbulence layer , while in the latter , particles either cool in the undecayed region of short extent , if their lorentz is sufficiently large , or in the background magnetic field beyond the microturbulent layer otherwise , in the absence of inverse compton losses that is .
this implies in particular that there is no well defined cooling lorentz factor for the microturbulence . secondly ,
as the turbulence decays rapidly , most of the synchrotron power is emitted in the region of largest magnetic power , hence the flux @xmath187 associated to the microturbulent layer peaks at @xmath164 . at times
such that @xmath154 , cooling in the background shock compressed field leads to the emergence of a secondary synchrotron component on top of the former , and one may now define a cooling lorentz factor in the background compressed field . at frequency @xmath148 ,
the ratio between these two components is of order @xmath188 , at the benefit of the latter . at the exit of the microturbulent layer
, the maximal lorentz factor can not exceed @xmath112 , so that the secondary component cuts - off at most at @xmath189 $ ] , which falls short of the gev range , see the discussion in sec .
[ sec : accel ] . in practice
, this suggests that most of the low energy emission in the optical and x - ray domains result from cooling in the background field , while the highest energy emission can be attributed to the presence of the microturbulence .
, now comparing the synchrotron spectrum in a homogeneous turbulence of strength @xmath174 ( top red curve ) , with a decaying microturbulence such that @xmath190 , @xmath191 ( bottom blue curve ) . for both ,
the blast parameters are @xmath175 , @xmath172 and the injection distribution index @xmath173 .
the orange dashed line represents the secondary synchrotron component associated to cooling in the background shock compressed field ( here @xmath192 g ) , which emerges on top of the microturbulent component because @xmath193 ( corresponding to case 2 of fig .
[ fig : spec_ainfm42pp1 ] ) .
the regime is slow cooling .
[ fig : spec2 ] , scaledwidth=49.0% ] figure [ fig : spec2 ] presents a concrete example of a spectral energy distribution , comparing the standard prediction for a homogeneous turbulence with @xmath174 ( top red line ) to a time evolving microturbulence with @xmath190 , also starting at @xmath174 ( bottom blue curve ) .
the microturbulent layer is such that @xmath194 , corresponding for instance to @xmath74 , @xmath195 . the upstream magnetic field @xmath192 g and as in fig .
[ fig : spec1 ] , @xmath172 , @xmath175 , @xmath196s , @xmath177 , which implies in particular that @xmath193 .
therefore , a secondary synchrotron associated to cooling in @xmath152 emerges on top of the microturbulent component .
the corresponding characteristic frequencies for the microturbulent component are @xmath197hz , @xmath198hz and @xmath199hz ; the cut - off frequency of the secondary synchrotron component @xmath200hz . in such a scenario ,
the microturbulent component would dominate in the x - ray and at higher energies , while the secondary component dominates in the optical ; at later times , the secondary component would come to dominate as well in the x - ray .
regarding the spectro - temporal evolution of the flux , one recovers features similar to those discussed above in the case @xmath132 .
in particular , the slow cooling slopes @xmath201 and @xmath202 remain unaffected , but the temporal index is different when @xmath156 , and for the fast cooling part of the electron population , both spectral and temporal indices are affected by @xmath65 .
the detailed values of @xmath3 and @xmath4 are given in table [ tab : fastnoic ] in appendix [ sec : appfnu ] .
see also figs .
[ fig : spec_ainfm42pp1 ] and [ fig : spec_asupm42pp1 ] for an illustration of the spectral shapes of the synchrotron spectra .
accounting for inverse compton losses modifies of course the cooling history of the particle .
the importance of inverse compton losses is generally quantified through the @xmath110 compton parameter , which may be written in a first approximation ( e.g. sari & esin 2001 , wang et al . 2010 ) as @xmath203\,\simeq\ , \frac{\epsilon_e}{\epsilon_b}\frac{\nu f_\nu\left[\nu_{\rm kn}(\gamma)\right ] } { \nu f_\nu\left(\nu_{\rm peak}\right ) } f_{\rm cool}\ , \ ] ] where the notation @xmath204 indicates that @xmath110 in general depends on the lorentz factor of the particle , due to klein - nishina effects .
these latter are quantified through the ratio of @xmath205 $ ] , i.e. the synchrotron flux at the frequency at which klein - nishina suppression becomes effective , to the peak of the synchrotron flux , @xmath206 .
finally , the factor @xmath207 denotes the fraction of cooling electrons , @xmath208 for fast cooling , @xmath209 for slow cooling .
the double dependence of @xmath110 on lorentz factor @xmath210 and distance to the shock front ( through @xmath70 ) that arises in the present model renders this problem quite complex . to simplify this task , it is assumed here that inverse compton losses dominate everywhere throughout the blast
this would occur for instance , if @xmath211 in the undecayed part of the microturbulent layer ( a generic assumption ) , if @xmath207 does not lie too far below unity , and at energies such that klein - nishina effects can be neglected . under such conditions
, one may expect @xmath212 everywhere else , because the magnetic field decays away from the shock .
, now comparing the synchrotron spectrum in a homogeneous turbulence of strength @xmath174 ( top red curve ) including inverse compton losses with @xmath213 , with a decaying microturbulence such that @xmath214 , @xmath170 , also including inverse compton losses with @xmath215 ( bottom blue curve ) . for both ,
the blast parameters are @xmath175 , @xmath172 and the injection distribution index @xmath173 . here
@xmath216 ( corresponding to case 1 of fig .
[ fig : spec_icasupm42pp1 ] ) .
the regime is slow cooling .
[ fig : spec3 ] , scaledwidth=49.0% ] in the presence of inverse compton losses , one can always define a cooling lorentz factor @xmath147 irrespectively of the value of @xmath65 . as the cooling history of the particles
is modified , so are the spectral slopes in the fast cooling part if @xmath217 , and also at low frequencies if @xmath218 . in this latter situation
indeed , the synchrotron spectrum is dominated at high frequencies by a slow cooling spectrum in the undecayed microturbulence , with a low energy extension from @xmath148 to @xmath219 [ see eq .
[ eq : numdbm ] ] with slope @xmath220 . at low frequencies ,
a secondary component associated to cooling in the background compressed field emerges if both @xmath156 and @xmath221 . the synchrotron spectra for @xmath132 share features similar to those discussed in sec .
[ sec : gradnoic ] , up to the modifications of the spectro - temporal indices .
one noteworthy distinction is the fact that the fast cooling spectral index has become @xmath222 , which may become larger than @xmath79 if @xmath223 , in which case most of the power lies at @xmath164 .
this may be attributed to a scaling of the compton parameter with frequency : higher frequencies correspond to higher lorentz factors , hence to cooling in a region of higher magnetic field , where the compton parameter is smaller , which implies that a smaller fraction of the particles energy is channeled into the inverse compton component .
figure [ fig : spec3 ] presents a concrete example of a spectral energy distribution , comparing the standard prediction for a homogeneous turbulence with @xmath174 including inverse compton losses with @xmath213 ( top red line ) to a time evolving microturbulence with @xmath214 , also starting at @xmath174 , with @xmath170 and @xmath215 ( bottom blue curve ) . for this case ,
@xmath216 , corresponding to case 1 of fig . [ fig : spec_icasupm42pp1 ] ; other parameters remain unchanged compared to previous figures .
the corresponding characteristic frequencies for the microturbulent component are @xmath224hz , @xmath225hz and @xmath226hz .
the spectro - temporal indices @xmath3 and @xmath4 are given in table [ tab : gradic ] for @xmath217 and [ tab : rapic ] in the opposite limit @xmath218 .
the generic spectral shapes are illustrated in figs .
[ fig : spec_icasupm42pp1 ] and [ fig : spec_icainfm42pp1 ] for these two cases , respectively . to compute the synchrotron spectral power of a relativistic blast wave ,
one usually solves a steady state transport equation for the blast integrated electron distribution and folds this distribution with the individual synchrotron power ( sari et al .
this method may be generalized to the present case by including an explicit spatial transport term and solving the transport equation through the method of characteristics .
it appears however more convenient to proceed with an approximation of the calculation of gruzinov & waxman ( 1999 ) , which integrates over the cooling history of each electron in the blast .
the present approximation neglects the integral over the angular coordinates ( 1d hypothesis ) , it neglects the hydrodynamical profile of the blast ( as in sari et al .
1998 ) and it neglects the time evolution of the blast physical characteristics over the cooling history of a freshly accelerated electron .
this reduces the integral to a more tractable form , with a closed expression , at the price of reasonable approximations .
the angular part of the integral indeed leads to a smoothing of the temporal profile of emission which may be re - introduced at a later time , see panaitescu & kumar ( 2000 ) .
the blast evolves on hydrodynamical timescales that are longer , or much longer than the radiative timescales .
the synchrotron spectral power of the blast is written as discussed in section [ sec : radmu ] , @xmath444 where @xmath138 denotes the differential per lorentz factor interval of the number of electrons swept by the shock wave by unit time .
it is defined in eqs .
[ eq : dnedg],[eq : dnedt ] . for commodity ,
the spectral power @xmath445 is written in the circumbust medium rest frame at redshift @xmath20 , whence the beaming factor @xmath446 , and all frequencies are written in the observer rest frame .
this means in particular , that the integrands are defined in the comoving blast frame , except @xmath447 of course , which is an observer frame quantity .
the received spectral flux @xmath145 then reads @xmath448 with @xmath449 the luminosity distance to the blast . the time integral in eq .
[ eq : appdfnu ] integrates the synchrotron power along the particle trajectory in the blast , i.e. from shock entry until the particle reaches the back of the blast . as it ignores the spatial profile of the blast , the upper bound @xmath135 is defined up to a factor of unity .
such prefactors of order unity are ignored in the following calculations , which furthermore always approximate down to broken powerlaws .
a correct overall normalization is provided at the end of the calculation . the spectral power radiated by an electron at time @xmath47 , of lorentz factor @xmath108
is approximated as @xmath451 with @xmath452\ .\ ] ] the peak frequency @xmath150 for lorentz factor @xmath108 and magnetic field @xmath70 is defined as ( e.g. wijers & galama 1999 ) @xmath453\,\equiv\ , \frac{3x_p}{2\pi}\frac{e\,\delta b}{m_e c } \gamma_e^2\frac{4\gamma_{\rm b}}{3(1+z)}\ , \label{eq : nusyn}\ ] ] with @xmath454 . for most cases , one might further approximate eq .
( [ eq : esyn ] ) as a delta function peaked around @xmath455 , but the low energy part in @xmath456 does actually play a role in several specific limits .
the additional factor of @xmath457 in eq .
( [ eq : esyn ] ) ensures proper normalization to the synchrotron energy loss rate integrated over frequency .
the physics of diffusive synchrotron radiation has received a lot of attention lately , notably because it might lead to distortions of the above spectral shape below and above @xmath150 ( see e.g. medvedev 2000 , fleishman & urtiev 2010 , kirk & reville 2010 , reville & kirk 2010 , mao & wang 2011 , medvedev et al .
the magnitude of such distortions is characterized by the wiggler parameter @xmath458 : @xmath459 if @xmath460 , the particle suffers a deflection of order unity in each cell of coherence of the turbulent field , hence the standard synchrotron regime applies . at the opposite extreme ,
if @xmath461 , the deflection per coherence cell does not exceed the emission angle of @xmath462 .
this leads to diffusive synchrotron radiation , with different slopes at low and high frequencies . in the intermediate regime , @xmath463 ,
the spectrum remains synchrotron like with some departures at @xmath464 and @xmath465 ( medvedev et al .
2011 ) . in the present scenario
, one can define a quantity @xmath466 immediately behind the shock front , in terms of @xmath53 and @xmath48 , @xmath467 with @xmath468 and @xmath469 the minimum lorentz factor of the electron population . for generic parameters , one thus finds @xmath470 or @xmath471 , but in any case , diffusive synchrotron effects can be neglected close to a highly relativistic shock front .
as the turbulence decays , its coherence length evolves , hence @xmath458 evolves in a non - trivial way , starting from @xmath466 . by the time @xmath128 , i.e. when the turbulence has relaxed to @xmath152 , one finds @xmath472 with @xmath473 , one can check that this value remains much larger than unity for generic grb blast wave characteristics .
considering for instance the example @xmath474 ( fast decay of the turbulence ) , @xmath280 , one finds @xmath475 .
diffusive synchrotron effects become important at late times when the blast wave has become moderately relativistic , but for the present purposes they can be neglected ; this justifies the above choice eq .
[ eq : esyn ] for @xmath476 .
the influence of synchrotron self - absorption is discussed briefly in sec .
[ sec : synch - self ] .
it does not affect the synchrotron spectra around and above the optical range but it may lead to interesting time signatures at frequencies below a ghz .
if @xmath69 , the particle cools gradually in the decaying microturbulent field , see eq .
[ eq : cool2 ] .
this section ignores the possibility of inverse compton losses , the effects of which are discussed in sec .
[ sec : appicgrad ] .
one then defines the critical frequencies @xmath477\ , \ ] ] which is associated to particles of lorentz factor @xmath112 and magnetic strength @xmath53 , and its counterpart @xmath478\ .\ ] ] the cooling lorentz factor enters the calculation through the upper bound on the integral defined in eq .
( [ eq : appdfnu ] ) : @xmath479 denotes as usual that which allows cooling on a timescale @xmath135 . to calculate @xmath147 ,
one first define the usual @xmath480/t_{\rm dyn}$ ] , which corresponds to the cooling lorentz factor in a homogeneous turbulence of strength @xmath53 .
the generic notation @xmath481 $ ] refers to the synchrotron loss time for a particle of lorentz factor @xmath108 in a magnetic field of strength @xmath70 .
then , if @xmath482 , cooling indeed occurs inside the undecayed part of the microturbulence so that @xmath483 . if @xmath154 , cooling instead occurs in the background field @xmath152 , meaning @xmath484 . in the intermediate regime @xmath485 , cooling takes place in the decaying part of the turbulence .
the cooling lorentz factor is then defined as the solution of @xmath486=t_{\rm dyn}\ , \ ] ] which leads to @xmath487 to summarize , @xmath488 the characteristic cooling frequency @xmath489\ , \label{eq : nuc}\ ] ] with @xmath490 the field @xmath491 corresponds to the field strength at the point at which particles of lorentz factor @xmath147 cool through synchrotron radiation .
the latter expression means that if @xmath156 , @xmath492 is given by the value of @xmath70 at time @xmath135 , while if @xmath156 , @xmath493 .
the factor @xmath494 can also be written as @xmath495 , with @xmath496 note that the value @xmath491 can also be understood as the smallest value of the magnetic field in the blast .
the characteristic frequency @xmath148 associated to particles of lorentz factor @xmath139 can also be derived similarly : @xmath497\ ,
\label{eq : num}\ ] ] with @xmath498 in short , this implies that @xmath151 is determined by @xmath491 if @xmath157 , and by @xmath499 otherwise , if the @xmath139 particles can cool in the decaying layer . to understand the former value
, one should note that for @xmath132 , most of the power of non - cooling particles ( @xmath500 ) is generated at the back of the blast , since the time integrated power @xmath501 .
, neglecting inverse compton losses , with the spectral indices as indicated .
case 1 : slow cooling scenario ( @xmath502 ) with @xmath156 , in which case the turbulence has not had time to relax to the background value @xmath152 ; case 2 : slow cooling scenario with @xmath154 ; case 3 : fast cooling scenario with @xmath503 ; case 4 : fast cooling scenario with @xmath504 ; case 5 : fast cooling scenario with @xmath505 . see the discussion in sec .
[ sec : appnoicasupm1 ] for the precise definitions of the various frequencies and for modifications to these spectra if @xmath146 or @xmath148 exceeds @xmath164 , or if @xmath506 is taken into account .
[ fig : spec_asupm1 ] , scaledwidth=49.0% ] depending on how @xmath146 , @xmath148 , @xmath165 and @xmath164 are ordered one relatively to each other , one obtains different synchrotron spectra .
the various possible shapes of these spectra with their characteristic indices are summarized in the five cases depicted in fig .
[ fig : spec_asupm1 ] and the spectro - temporal indices of @xmath507 are provided in table [ tab : gradnoic ] .
this table assumes a decelerating blast wave in an external density profile @xmath508 . at very low frequencies ,
the synchrotron spectrum may be modified by opacity effects .
the absorption break frequency generally lies well below the optical domain , e.g. for a homogeneous turbulence , @xmath509 ( constant density profile ) and a slow cooling regime , one finds @xmath510 ( see panaitescu & kumar 2000 ) . in this standard case ,
the frequency index of @xmath145 is @xmath511 below the absorption frequency , e.g. granot et al .
( 1999 ) and references therein .
accounting for the decaying microturbulence modifies the situation as follows .
all calculations are performed in the comoving downstream frame , as indicated by the primes .
moreover , the self - similar profile of the blast and the secular evolution of the blast characteristics are neglected , as in the rest of app . [
sec : appfnu ] .
one then defines the synchrotron self - absorption coefficient ( rybicki & lightman , 1979 ) : @xmath512 with @xmath513 the spectral synchrotron power emitted per electron per frequency interval , as defined in eq .
[ eq : esyn ] up to the change of frame for the frequency .
here however , one considers only the low frequency part @xmath514 since one is interested in frequencies well below @xmath148 .
the electron distribution function @xmath515 depends on the distance to the shock front when cooling is efficient . since
only the minimum lorentz factor is of importance in the calculations of synchrotron self - absorption , one can approximate the distribution as a unique powerlaw , @xmath516 with @xmath517 at distances such that cooling is inefficient for particles of lorentz factor @xmath139 , and @xmath518 position dependent at larger distances from the shock front . of course , in a slow cooling regime , @xmath517 everywhere in the blast .
the index @xmath519 departs from @xmath520 only when cooling becomes efficient ; its exact value does not affect the following up to factors of order unity .
the above integral leads to @xmath521 with @xmath522 $ ] .
this local absorption coefficient depends on position both through @xmath518 ( when cooling is efficient ) and through @xmath70 , in the presence of decaying turbulence .
one may thus rewrite the absorption coefficient as @xmath523 with @xmath524 the absorption coefficient defined according to eq .
[ eq : ap2 ] with @xmath525 and @xmath526 .
this formulation allows to write @xmath527 as a broken power law function of the distance to the shock front .
the specific intensity along a ray path obeys the radiative transfer equation @xmath533 with the usual formal solution @xmath534 here @xmath535 , @xmath536 denotes the distance from the shock front , and @xmath537 by convention .
the parameter @xmath538 consequently represents the total optical depth of the blast which can be derived by integrating the broken power law form of eq .
[ eq : ap3 ] over @xmath100 , from @xmath539 up to @xmath540 .
the above formulation of the solution of the equation of radiative transfer is particularly appealing because it leads to a simple evaluation of @xmath541 in various cases .
in particular , if the regime is slow cooling , @xmath542 everywhere and @xmath543 becomes uniform in the blast , as in the homogeneous ( slow cooling ) case .
then one finds @xmath544 in the optically thick limit @xmath545 , the ratio @xmath546 leads to the standard slope @xmath511 for @xmath145 below the absorption frequency @xmath547 , which is defined by the condition @xmath548 at @xmath549 . note that @xmath538 shares with the absorption coefficient @xmath527 the scaling @xmath550 . in the optically thin limit @xmath551 , meaning @xmath552 , one recovers the slope @xmath201 since @xmath553 .
the non trivial dependence of @xmath547 on time leads to a non trivial time dependence of the flux below the break frequency .
this dependence is not discussed here , but can be calculated through @xmath538 .
the fast cooling regime leads to a more complicated shape of the spectrum in the transition region between the asymptotic thin and thick regimes ( for a similar discussion with a homogeneous turbulence , see granot et al .
one needs to define here an intermediate optical depth , @xmath554 with @xmath555 the location at which particles of lorentz factor @xmath139 start to cool .
the fast cooling regime implies that by the back of the blast , such particles have cooled down to @xmath147 .
one now defines an absorption frequency @xmath556 such that @xmath557 at @xmath558 . because the optical depth is an increasing function of distance to the shock front , @xmath559 ( possibly @xmath560 ) and @xmath561 .
therefore , at frequencies @xmath562 , the shell is optically thick for @xmath563 and the standard synchrotron self - absorption spectrum emerges . at frequencies @xmath564 , the whole shell is optically thin to synchrotron radiation , so that the standard @xmath514 spectrum calculated above applies .
however , for @xmath565 , opacity competes with local emission at distances @xmath566 , and the overall spectrum departs from either asymptote .
the exact spectral shape can be derived from eq .
[ eq : inu ] , using the calculated value of @xmath538 .
most of the discussion so far has ignored the characteristic frequency associated to the maximal energy of the acceleration process . yet , the production of gev photons through synchrotron radiation of electrons already push ultra - relativistic fermi acceleration to its limits , as discussed e.g. in piran & nakar ( 2010 ) , kirk & reville ( 2010 ) , barniol - duran & kumar ( 2010 , 2011a ) , lemoine & pelletier ( 2011c ) , sagi & nakar ( 2012 ) , bykov et al .
these studies assume a homogeneous downstream turbulence , with either microscale high power turbulence or simply a shock compressed magnetic field .
section [ sec : accel ] extends these calculations to an evolving microturbulence , which brings in further constraints and new phenomena . a brief comparison with the light curve of grb090510 , which so far provides the earliest follow - up in the optical through gev ,
is then proposed .
the maximal energy is determined through the comparison of the acceleration timescale @xmath227 to other relevant timescales , e.g. the energy loss and dynamical timescales . given that the energy gain per fermi cycle is of order 2 ( gallant & achterberg 1999 , achterberg et al .
2001 , lemoine & pelletier 2003 ) , it suffices to compare in each respective rest frame the downstream @xmath228 and the upstream @xmath229 residence times to the other timescales . the upstream residence time @xmath230 ( lemoine & pelletier 2003 ) assuming that the particle only interacts with the background field where @xmath231 and @xmath108 represents the particle lorentz factor in the downstream frame , @xmath232 . considering this residence time leads to a conservative limit on the maximal energy , because scattering in the microturbulent field seeded in the shock precursor should also play a role , see plotnikov et al .
( 2012 ) for further discussion on this issue . comparing the above residence time to the age of the shock wave
@xmath43 then leads to a maximal lorentz factor ( downstream frame ) @xmath233 depending on the dynamics of the microturbulence downstream , this maximal lorentz leads to to photons of energy @xmath234 through radiation in the @xmath53 field , or @xmath235 if the particle radiates in the background shock compressed field @xmath152 , as proposed in barniol - duran & kumar ( 2010 , 2011a ) .
this latter case applies in particular if @xmath81 , @xmath236 and if downstream inverse compton losses can be neglected .
if however , @xmath237 , then the particle radiates in the stronger @xmath53 , while if @xmath132 and @xmath238 , the particle radiates in a field of strength @xmath239 , which for these high energies lies in practice close to @xmath53 .
for instance , if @xmath162 , one finds @xmath240 equation [ eq : emaxabd ] matches the `` confinement '' estimates of piran & nakar ( 2010 ) , barniol - duran & kumar ( 2011a ) , who considered only a shock compressed background field in the downstream , while the other estimates eqs .
[ eq : emaxadb ] , [ eq : emaxadbi ] hold in the present more realistic setting because of the high value of magnetic field behind the shock front .
note that the particle is actually confined close to the shock front , because in the ultra - relativistic regime , the shock front never lags further than @xmath241 behind the particle ; escape can only take place through the boundaries , but this would lead to a less stringent condition on maximal energy than the above age constraint , see e.g. bykov et al .
( 2012 ) .
the above suggests that the upstream magnetic field can lie well below @xmath242 g and yet lead to the production of multi - gev photons because the maximal lorentz factor particles cool in a strong field , close to the microturbulent value @xmath53 behind the shock .
furthermore , if one neglects altogether the upstream magnetic field , the discussion in plotnikov et al .
( 2012 ) suggests that scattering the microturbulent field generated in the shock precursor may lead to lorentz factors sufficiently large to produce gev photons through synchrotron radiation in @xmath53 .
li & waxman ( 2006 ) and li & zhao ( 2011 ) have argued that the comparison of the upstream residence time with the timescale for inverse compton losses imposes a stringent lower bound on the upstream magnetic field , suggesting an efficient amplification of the latter .
more specifically , li & zhao ( 2011 ) find that @xmath243 ( with @xmath244 denoting the total magnetic field upstream of the blast ) is imposed by the synchrotron model of @xmath245lat bursts with extended gev emission .
these bounds have been challenged in barniol - duran & kumar ( 2011b ) and sagi & nakar ( 2012 ) .
nevertheless , such bounds agree well with the general picture that microinstabilities self - generate a microturbulent field in the shock precursor , which is then transmitted downstream and there forms the microturbulent layer .
one expects that in the shock precursor , the microturbulent field @xmath246 ( as measured in the upstream frame ) , in which case one finds @xmath247 .
this field satisfies the constraints of li & zhao ( 2011 ) .
consider now the situation downstream , which leads to much more severe contraints on @xmath248 through the comparison of @xmath249 with the timescale of energy loss . out of commodity and simplicity ,
the limit on the maximal energy is often quoted with a bohm estimate @xmath250 , with @xmath251 in the downstream frame . ignoring inverse compton losses ,
this leads to @xmath252 assuming for the moment that @xmath70 is homogeneous , with strength characterized by @xmath253 .
this leads to @xmath254 which matches the estimate of piran & nakar ( 2010 ) up to a factor 2 .
however , this estimate must be corrected for two effects : ( 1 ) the bohm approximation likely breaks down at these large lorentz factors ; ( 2 ) the evolution of the microturbulence away from the shock front modifies the scattering rate of the particles .
the bohm approximation for @xmath249 fails at the maximal lorentz factors because @xmath255 for a lorentz factor given by eq .
[ eq : gmaxb ] .
this implies that the particle only suffers a random small angle deflection of order @xmath256 as it crosses a coherence cell of size @xmath257 , hence the residence timescale is rather given as : @xmath258 , much larger than the above bohm estimate .
this of course reduces significantly the maximal energy of photons ( kirk & reville 2010 , lemoine & pelletier 2011c , plotnikov et al .
2012 ) . at the present time
, one can not exclude that the spectrum of turbulent modes extends to wavenumbers @xmath259 that would allow gyroresonant interactions in the downstream ; whether or not this happens depends on the physics of instabilities in the far upstream , which are still debated .
were this the case , the bohm estimate should nevertheless be corrected by a factor that accounts for the diminished magnetic power at those resonant scales , compared to the maximum power at @xmath260 . in order to obtain a conservative estimate of the maximal energy , one should therefore rely on the above @xmath261 . comparing this residence timescale with the synchrotron loss time , this would lead to a maximal lorentz factor @xmath262 corresponding to a maximum photon energy @xmath263 _ provided the radiation takes place in the microturbulent field of strength @xmath53_. if radiation were to take place in the background shock compressed field @xmath152 , the maximal photon energy would rather be @xmath264 however , once the evolution of the microturbulence is accounted for , these estimates are modified towards more optimistic values .
consider first the scaling of @xmath265 , with @xmath266 now designing the gyroradius in the local magnetic field , and @xmath29 the coherence length of this local magnetic field .
then @xmath267 with @xmath94 , the following qualitative picture emerges .
if the turbulence decays gradually , meaning @xmath268 , the coherence length increases more rapidly away from the shock front than the magnetic field loses power , so that the scattering timescale tends towards a bohm scaling ( albeit , in a weaker magnetic field ) as time increases ( see also katz et al .
2007 for a similar picture ) .
one then obtains ( see further below ) maximal energy estimates close to the bohm scaling .
if , however , the turbulence decays fast , meaning @xmath269 , then the scattering timescale increases with time , possibly faster than @xmath47 if @xmath270 , in which case scattering becomes impossible in the decaying part of the microturbulent layer .
furthermore , mirror like interactions in the background shock compressed magnetic field , which lies transverse to the shock normal , do not allow repeated returns to the shock front ( lemoine et al .
2006 , see also sec .
[ sec : fcons ] ) .
the particles must then scatter in the undecayed microturbulent field , if @xmath269 , in which case the maximal energy is then determined by the comparison of @xmath249 to @xmath144 .
these two scenarios are examined in turn .
assume first @xmath271 , meaning more specifically @xmath272 .
then the small angle scattering timescale @xmath273 , which must be compared to the synchrotron loss time @xmath274 , where the last factor accounts for the evolution of the magnetic field strength with time , outside the undecayed layer .
this leads to @xmath275^{-(1+\alpha_t)/(3 + 3\alpha_t + 1/\alpha_\lambda)}\ .\ ] ] the term within the brackets can be rewritten in a more compact way as @xmath276 , with @xmath277 the classical electron radius , so that in the limit @xmath278 , one recovers the estimate @xmath279 in a homogeneous microturbulence .
the fiducial values @xmath162 and @xmath280 discussed previously lead however to @xmath281 which , through synchrotron radiation in a magnetic field of strength @xmath282 leads to photons of typical energy @xmath283 one should stress that this result depends on the particular value of @xmath65 and @xmath64 , but for the above fiducial values , it remains close to the bohm estimate and allows the production of gev photons for conservative assumptions .
note finally that the stretching of the coherence length with time does not imply that the particles of lorentz factor @xmath284 suffer gyroresonant interactions away from the shock , nor does it require that the turbulent power spectrum extends over many decades . for the above fiducial values ,
the particles interact with modes of wavelength @xmath285 . in a fast decaying scenario , meaning more exactly @xmath286 , scattering must take place in the undecayed part of the microturbulent layer , otherwise the particles would not see the microturbulent layer and fermi acceleration would not take place . the comparison @xmath287 leads to @xmath288 which would lead to @xmath289mev photons only , even through radiation in a field as strong as @xmath53 . to reach the gev range , ceteris paribus
, one needs to increase the spatial extent of the undecayed microturbulence , i.e. to increase @xmath290 by @xmath291 orders of magnitude . for @xmath292 , indeed , @xmath293 , so that the particle both scatters and cools in the undecayed part of the microturbulent layer . to summarize , the above analysis of scattering and maximal photon energies leads to the following qualitative picture .
if the turbulence decays gradually , the particles scatter while the microturbulence decays and the maximal energy of the photons typically falls in the gev range for fiducial parameters and rather conservative estimates of the scattering properties .
if , however , the turbulence decays rapidly , gev photons can only be produced if the undecayed part of the turbulent layer extends far enough to accomodate both the scattering and the cooling of the maximal energy electrons in that layer .
this latter requirement places strong constraint on the parameter @xmath120 that characterizes for how long the undecayed turbulence can survive .
requisite values can be obtained if , for instance , @xmath294 and @xmath295 , which can not be excluded at present . finally , it has also been noted that the constraints on the upstream residence time are much weaker if the maximal energy electrons can radiate in a field close to @xmath296 . in practice , this implies that the value of the upstream magnetic field is not well constrained .
this alleviates the apparent strong magnetization problem discussed in he et al .
( 2011 ) .
section [ sec : synchspec ] together with appendix [ sec : appfnu ] have emphasized the differences between the standard synchrotron spectra for homogeneous turbulence and those calculated with the account of decay of the microturbulence .
this section confronts such signatures with the observational data on grb090510 , which so far has provided the earliest follow - up on a broad spectral range , with near simultaneous detection in the optical , x range and gev range ( de pasquale et al .
2010 ; ukwatta et al . 2009 ; guiriec et al .
2009 ; ackermann et al .
2010 ; longo et al . 2009 ; hoversten et al . 2009 ; golenetskii et al .
2009 ; ohmori et al .
this choice is further motivated by the analyses of barniol - duran & kumar ( 2009 , 2010 , 2011a ) and he et al .
( 2011 ) , who have argued that the extended emission , from the optical to the gev range is compatible with an afterglow spectrum calculated without magnetic field generation beyond direct shock compression of the interstellar field .
one may also interpret the data with a high magnetization of the blast , closer to the traditional estimates of @xmath253 , although it then requires an extraordinarily low upstream density @xmath297@xmath10 ( e.g. de pasquale et al .
2009 , corsi et al .
2010 ) ; this possibility is not considered here .
the model of barniol - duran & kumar ( 2009 , 2010 , 2011a ) then suggests that magnetic field generation processes are ineffective , and that this burst provides the closest relative to a `` clean '' relativistic blast wave in a weakly magnetized medium as discussed in the previous sections , one should expect that the gev photons have been produced in the microturbulent layer rather than in the background magnetic field .
however , lower energy photons in the optical and in the x range are typically produced by particles of lorentz @xmath139 , which lose energy on a much longer timescale than the particles of lorentz factor @xmath298 , hence in a weaker magnetic field , possibly the background shock compressed value @xmath152 . in any case the microturbulence is to play a key role in shaping the spectra from the gev down to lower energies .
this motivates further the search for a signature of this microturbulence .
unfortunately , the comparison is not straightforward , because of the diversity and complexity of the synchrotron spectra .
the quantities that characterize the microturbulence appear as new free parameters from the point of view of phenomenology , hence they enlarge the dimensionality of parameter space and add degeneracy .
for this reason , the following does not attempt to fit accurately the light curves in the gev , x - ray and optical , but rather accomodates the different spectro - temporal indices in the different time domains of the observations .
the key observation of the barniol - duran & kumar ( 2009 , 2010 , 2011a ) interpretation is that the spectrum is produced in a slow cooling regime , and that in the case of grb090510 , @xmath148 transits across the optical at @xmath299s , in order to produce the observed break .
once this condition is satisfied , and provided spectro - temporal indices match the observed values , one obtains a satisfactory fit to the light curves ( as it has been checked ) .
the following discussion reveals that the current data do not allow one to pick out a preferred model , meaning conversely that it is possible to reproduce the salient features of the barniol - duran & kumar ( 2009 , 2010 , 2011a ) model with concrete characterizations of the microturbulence .
each model leads to rather specific signatures that might be identified with further higher accuracy and earlier observations .
some of these signatures are pointed out in the following , although more work is needed to define an observation strategy capable of pinpointing the characteristics of the microturbulence .
grb090510 is a short burst at redshift @xmath300 whose light curve is characterized as follows : emission is seen in the lat range up to @xmath301s , with spectro - temporal indices @xmath302 , @xmath303 ; x - ray and optical follow - up start at about this time ; from @xmath301s to about @xmath304s , @xmath305 , increasing to @xmath306 afterwards , while @xmath307 ; @xmath308 before @xmath299s , then @xmath309 afterwards . the late time evolution beyond @xmath299s
is more difficult to reproduce , in particular the steepening of the x - ray lightcurve , which is attributed by he et al .
( 2011 ) to sideways expansion , but which does not match the apparent shallower decay in the optical up to @xmath310s ( nicuesa guelbenzu et al .
2012 ) . to compare this light curve with the afterglow computed in app .
[ sec : appfnu ] , consider first the possiblity of @xmath132 ( gradual decay ) without inverse compton losses .
the latter assumption has been discussed at length in he et al .
( 2011 ) . as shown in fig .
[ fig : spec_asupm1 ] , the spectral shape of the synchrotron spectrum in the slow cooling limit with @xmath132 retains the same form as the standard afterglow , but the values of the characteristic frequencies and therefore the time dependences are modified .
therefore , @xmath311 if @xmath312 as observed . in the model of barniol - duran & kumar ( 2010 ) ,
( 2011 ) , @xmath146 lies below the lat range , so that @xmath313 which takes values between @xmath314 and @xmath315 for @xmath173 and @xmath316 $ ] , in good agreement with the data , although the influence of @xmath65 is too weak to provide significant constraints .
the temporal slope in the gev range also depends mildly on @xmath65 , taking values from @xmath317 at @xmath318 ( @xmath173 ) to @xmath319 at @xmath320 ( @xmath173 ) .
it fits with the observations , given that the measured slope may be contaminated by prompt gev emission , see the discussion in he et al .
the best probes for @xmath65 remain the temporal slopes in the x and optical domains : @xmath3 takes values between @xmath321 and @xmath322 in the x range ( for @xmath173 ) , and between @xmath323 and @xmath324 ( @xmath173 ) in the optical when @xmath325 .
current data do not allow a precise determination of @xmath65 , but it may be noted that the fiducial value @xmath162 fits well the data . finally , requiring @xmath326ev at @xmath299s determines @xmath65 as a function of the other parameters .
for the fiducial values characterizing the turbulence , @xmath327@xmath10 and @xmath328 g , @xmath329 $ ] because the turbulence has not had time yet to relax to the background magnetic field @xmath330 .
then @xmath331 so that @xmath326ev at @xmath299s implies @xmath332 up to small logarithmic corrections .
it is somewhat remarkable that the value of @xmath65 falls close to the fiducial value discussed previously .
also , one may note that the upstream magnetic field does not enter eq .
[ eq : ata ] , because the particles actually cool in the decaying microturbulence .
one may verify that the above choice of parameters gives a light curve in good agreement with the observed data , except for the late time shallow decay in the optical mentioned above . to pursue this comparison , consider now the possibility that @xmath333 .
the fast decay of the microturbulence in this case actually mimics somewhat the scenario originally proposed by barniol - duran & kumar ( 2010 , 2011a ) : the particles get accelerated in a microturbulent layer behind the shock front but cool where the microturbulence has died away .
although , as discussed in some details in sec .
[ sec : accel ] , it appears necessary to require that the particles of lorentz factor @xmath298 actually cool in the microturbulent layer in order to produce the gev photons , in other words @xmath334 at @xmath335s , which implies @xmath336 or to put it more simply , that the undecayed part of the microturbulent layer with @xmath337 extends for some @xmath338 skin depths at least .
two different behaviors can be observed , depending on the ratio @xmath160 , i.e. whether the turbulence has relaxed down to @xmath152 , in which case a synchrotron component associated to @xmath152 would emerge on top of the microturbulent synchrotron spectrum .
the maximum value of @xmath160 is obtained for @xmath339 : @xmath340 so that both possibilities have to be envisaged . if @xmath341 up to late times , and @xmath342 beyond @xmath299s , the optical flux should decay fast with @xmath343 , which disfavors this possibility . if now @xmath154 at some point , the synchrotron component produced by the cooling of particles in @xmath330 emerges and dominates over that produced by the microturbulence , at least at low frequencies ( optical , x ) .
this @xmath152 component cuts off at @xmath344 $ ] , as discussed in sec .
[ sec : fastnoic ] ; @xmath345 increases with time , and for reasonable choices of parameters , it lies above the x range . at the time at which @xmath346 , however , the temporal decay slope in the x range changes from @xmath347 ( due to cooling in the microturbulence ) to @xmath348 , which implies a change from steep to shallow , which is not observed .
therefore , this scenario is disfavored as well .
it is difficult at the present to rule it out clearly , as other effects , due to jet breaking for instance , could complicate the temporal scaling .
consider now the possibility @xmath349 , without inverse compton losses as above .
similar considerations regarding the maximal energy suggests that @xmath350 , implying @xmath292 .
the turbulence decays fast so that , for a reasonable choice of parameters , e.g. @xmath351 , @xmath352 , @xmath353 , the condition @xmath154 may be fulfilled at @xmath354s .
this implies that the background synchrotron component emerges on top of the microturbulent synchrotron spectrum , and dominates at low frequencies .
one then recovers the standard afterglow spectrum proposed by barniol - duran & kumar ( 2010 ) , except that the gev photons are produced in the microturbulent layer . in the gev range ,
the indices @xmath355 match the standard predictions for the fast cooling population and thus agree with the data ; the overall afterglow provides a satisfactory fit to the available data . in order to discriminate this model
, one would need to access the early light curve in the frequency range @xmath356 ( i.e. x - ray ) in which both the spectral and temporal indices evolves strongly with @xmath65 : @xmath355 go from @xmath357 to @xmath358 as @xmath65 goes from @xmath359 to @xmath360 . for reference , @xmath361 .
another interesting solution arises in this scenario when @xmath341 up to late times .
this may be realized if @xmath362 for instance .
then the frequency @xmath363 $ ] transits through the optical at @xmath299s if @xmath364 , as required . during the interval @xmath365s ,
the optical lies in the range @xmath366 with @xmath367 $ ] . to reconcile @xmath3 with @xmath368 ( nicuesa gulbenzu et al .
2012 ) , one would need to impose @xmath369 , while @xmath370 at that time rather requires @xmath371 .
thus @xmath372 provides a compromise , which is marginally acceptable by the data , that presents the clear advantage of a shallower decay in the optical than in the x - ray at a time at which it is observed . at @xmath373s , @xmath219 transits across the optical ( see its expression above ) , so that the optical and the x - ray domains decline alongside at later times .
whichever ratio @xmath160 is considered , the model with @xmath218 predicts a clear signature in the early optical lightcurve , with @xmath374 when @xmath375 , which leads to a mild growth the optical flux . de pasquale et al .
( 2010 ) report @xmath376 between @xmath377s and @xmath299s , while nicuesa gulbenzu et al . (
2012 ) rather indicate @xmath378 , therefore it is not possible to rule out or confirm this model at present .
consider finally the possiblity of strong inverse compton losses .
as before , it is assumed here that these losses dominate throughout the blast , although a more realistic model should include the possibility of weak inverse compton losses at the highest energies due to the klein - nishina suppression of the cross section .
nevertheless , one may constrain this scenario with the optical and x data , as follows .
the dominance of inverse compton losses for particles of lorentz factor @xmath139 implies that @xmath379 , since @xmath380 see eq .
[ eq : ym ] for the definition of @xmath381 .
consider first the possibility @xmath217 , this implies that @xmath382kev and @xmath383kev at @xmath384s , so the x range always fall in the fast cooling portion .
the spectral slope can not be @xmath385 in that range , otherwise it could not match the observed @xmath386 ; the slope must thus be @xmath222 , which fits @xmath387 only for @xmath388 . in turn , this would imply a positive value of @xmath389 if @xmath390 ( fast cooling ) , so the regime at time @xmath391s must be that of slow cooling ( case 1 or 2 depicted in fig .
[ fig : spec_icasupm42pp1 ] ) .
one should further require that @xmath148 goes into the optical at @xmath299s as before .
these features can be achieved for the typical fiducial values considered before , @xmath352 , @xmath392 , with @xmath393 .
an interesting feature of this scenario is that the ratio @xmath394 can take values smaller or larger than unity for standard parameters of the microturbulence and @xmath395 . as it depends sensitively on the value of the upstream magnetic field , as @xmath396 , the transition from @xmath156 ( case 1 in fig .
[ fig : spec_icasupm42pp1 ] ) to @xmath154 ( case 2 in fig .
[ fig : spec_icasupm42pp1 ] ) may well occur around @xmath397s .
once @xmath188 , the optical domain lies in the fast cooling region corresponding to @xmath398 and indices @xmath399\,\simeq\,(1.1 , 1.1)$ ] .
the optical domain then decays more slowly than the x domain , @xmath3 being in marginal agreement with the inferred value @xmath400 ( nicuesa gulbenzu et al .
note that the frequency @xmath165 increases in time as @xmath401 , but remains below the x domain if @xmath402 .
therefore the break at @xmath310s in the optical can not be attributed to another frequency crossing the optical domain ; it might however result from jet sideways expansion in that model . in order to probe and discriminate this model
, one would need to access the very early light curve , while it is in the fast cooling regime and measure the temporal decay slopes of the x and optical domains .
the above modelling has implicitly assumed a compton parameter @xmath403 ; one can check that for such a value , the inverse compton component of optical photons at @xmath8s remains below the synchrotron flux in the gev domain .
if @xmath218 , the spectra depicted in fig . [ fig : spec_icainfm42pp1 ] generally resemble at high energies a slow cooling spectrum in the microturbulent field .
since @xmath404kev at @xmath354s , the x domain always lies on the slow cooling portion of the microturbulent synchrotron spectrum with therefore correct values for @xmath405 , @xmath386 . among the 5 cases shown in fig .
[ fig : spec_icainfm42pp1 ] , the only two reasonable ones are 5 and 2 . indeed , case 1 is unlikely because it requires a small external density to achieve slow cooling , but then @xmath160 becomes large and case 2 actually applies .
similarly , case 4 is unlikely because @xmath406 as argued above .
case 3 leads to @xmath407 , which goes contrary to the observational data .
therefore , cases 5 and/or 2 remain as the viable alternatives .
however , one must now impose @xmath408kev at @xmath6s , so that the secondary synchrotron component associated to @xmath152 cuts off below the x range , otherwise the x domain would fall in the fast cooling part of that synchrotron component .
this constraint is fulfilled for the previous fiducial values @xmath402 , @xmath409 , @xmath410 and for @xmath411 .
interestingly , the optical falls on the fast cooling part of the secondary synchrotron component at @xmath412s , since @xmath413 , therefore it decays less fast than the x range , with @xmath3 in marginal agreement with the observed value , as previously . the temporal slope in the optical at very early times @xmath414s provides an interesting check of this scenario : it should be flat or even weakly decaying , corresponding to case 3 in fig . [
fig : spec_icainfm42pp1 ] , which represents the analog of case 5 when @xmath156 . finally , as discussed before
, one must require @xmath292 in order to produce gev photons directly in the undecayed part of the microturbulent layer . in this model ,
the microturbulent layer produces both the gev and x photons , while the optical photons are produced in the shock compressed background field .
according to the modelling of barniol - duran & kumar ( 2009 , 2010 , 2011a ) , the magnetization of the blast wave can take a large range of values , from @xmath415 to @xmath416 , depending on the external density .
the latter value is given in de pasquale et al .
( 2009 ) and corsi et al .
( 2010 ) , assuming very low upstream densities @xmath417@xmath10 .
the general trend is that , independently of @xmath418 , the downstream magnetic field appears to coincide with the shock compressed magnetic field @xmath419 , corresponding to @xmath420 g .
as noted in he et al .
( 2011 ) , this generally implies large upstream magnetizations , since @xmath421 if @xmath253 is calculated in terms of @xmath152 , whereas the typical interstellar magnetization is more of the order of @xmath422 . as discussed in the previous sections , accounting for a decaying microturbulent layer alleviates the need for a large value of @xmath423 , which allows to reconcile the apparent @xmath253 with a more standard upstream magnetization . nevertheless , it is interesting to ask what would happen if the upstream magnetization were truly high , e.g. if @xmath424 g and @xmath425@xmath10 . if @xmath426 is brought upwards by several orders of magnitude , then magnetization effects are to affect the shock physics .
for instance , the pic simulations of sironi & spitkovsky ( 2011 ) indicate that , for @xmath26 and @xmath427 , the shock is no longer mediated by the filamentation instability , but by the magnetic barrier associated to the ( transverse ) background magnetic field . as discussed in lemoine & pelletier ( 2010 , 2011a ) ,
such a transition occurs when the growth timescale of the filamentation instability becomes larger than the timescale on which upstream plasma elements cross the shock precursor .
this transition is predicted to occur when @xmath428 , with @xmath429 the fraction of shock energy carried by suprathermal particles . for @xmath430 ,
the limiting magnetization becomes @xmath431 , to be compared with @xmath432 in terms of the above fiducial values .
if @xmath433 , microturbulence can not be excited at the shock front , hence the downstream plasma must be permeated with the background shock compressed field only . however , one should not expect particle acceleration to proceed and indeed , the pic simulations of sironi & spitkovsky ( 2011 ) indicate no particle acceleration for @xmath427 , @xmath434 . in order to obtain particle acceleration , one needs to find a source of turbulence in the downstream plasma , that would be able to scatter particles back to the shock front faster than they are advected .
this possibility can not be discarded at the present time but it leads to some form of paradox : given that accelerated particles probe a length scale @xmath435 behind the shock if they interact with a magnetic field coherent over scales @xmath436 ( lemoine & revenu 2006 , lemoine et al .
2006 ) , the instability needs to produce large scale modes on spatial scales @xmath437 with a growth timescale @xmath438 .
the most reasonable scenario is thus to assume that microturbulence exists behind the shock , which restricts the possible values of the upstream magnetization . finally , if @xmath439 at some early time , but @xmath440 at some late time , because @xmath441 while @xmath426 remains constant , one should observe a very peculiar signature in the light curve due to the initial absence of an extended shock accelerated powerlaw ( lemoine & pelletier 2011b ) .
current understanding of the formation of weakly magnetized collisionless shocks implies the formation of an extended microturbulent layer , both upstream and downstream of the shock .
it is expected that the microturbulence behind the shock decays away on some hundreds of skin depth scales , and recent high performance particle in cell simulations have confirmed this picture . borrowing from these simulations , section [ sec : synchspec ] has described the decay of the microturbulence as a powerlaw in time , with an index that can take mild or pronounced negative values , depending on the breadth of the spectrum of magnetic perturbations seeded immediately behind the shock .
the synchrotron signal of a shock accelerated powerlaw of particles radiating in this evolving microturbulence has been calculated , and the results are reported in app .
[ sec : appfnu ] .
this calculation reveals a rather large diversity of possible signatures in the spectro - temporal evolution of the synchrotron flux of a decelerating relativistic blast wave , commonly parameterized under the form @xmath442 .
the diversity of signals is associated with the possible values of the ratio @xmath160 , which characterizes whether the blast has had time to relax to the background shock compressed field or not by the comoving dynamical time @xmath135 , as well as the different cooling regimes ( fast cooling or slow cooling on timescale @xmath443 ) and the influence or not of inverse compton losses on the cooling history of particles .
one point on which emphasis should be put , is that these signatures are potentially visible in early follow - up observations of gamma - ray bursts on a wide range of frequencies .
the detailed synchrotron shapes and indices @xmath3 , @xmath4 are discussed in app .
[ sec : appfnu ] .
the microturbulence controls the scattering process of particles during the acceleration stage and it therefore controls the maximal energy of shock accelerated particles .
detailed estimates for this maximal energy and for the corresponding energy of synchrotron photons have been provided .
it has been argued that the observation of gev synchrotron photons from gamma - ray burst afterglows implies either that the microturbulence decays rather slowly ( as actually observed so far in pic simulations ) , or that the spatial extent of the undecayed layer of microturbulence is large enough to accomodate both the scattering and the cooling of these particles . in the former case , the stretching of the coherence length of the microturbulence that accompanies the erosion of magnetic power improves the acceleration efficiency and brings it close to a bohm scaling of the scattering frequency .
finally , this paper has provided a concrete basis for the phenomenological models of barniol - duran & kumar ( 2009 , 2010 , 2011a ) , which interpret the extended gev emission of a fraction of gamma - ray bursts as synchrotron radiation of shock accelerated particles in the background shock compressed field .
microturbulence must play an important role in these models for several reasons : ( 1 ) it is expected to exist behind the shock front of a weakly magnetized relativistic shock wave ; ( 2 ) it ensures the development of fermi acceleration , which can not proceed without efficient scattering ; ( 3 ) it controls the production of photons of energy as high as a few gev . using the predictions of the spectro - temporal dependence of @xmath145 with a dynamical microturbulence
, one can reproduce satisfactorily the phenomenology of the models of barniol - duran & kumar ( 2009 , 2010 , 2011a ) , as examplified here with the particular case of grb090510 . in the present construction ,
the decaying microturbulent layer permits the generation of high energy photons close to the shock front , while the low energy photons are produced away from the shock , where the microturbulence has relaxed to ( or close to ) the background shock compressed field . in this setting , the broadband early follow - up observations of gamma - ray bursts afterglows open an exceptional window on the physics of collisionless shocks .
more work is certainly needed to study the phenomenology of evolving microturbulence and its relation with gamma - ray bursts light curves , in particular to define a proper observational strategy capable of inferring the characteristics of the microturbulence .
it is a pleasure to thank p. kumar , z. li , g. pelletier and x .- y . wang for useful discussions .
this work has been supported in part by the peps - pti program of the inp ( cnrs ) .
it has made use of the mathematica software and of the levelscheme scientific figure preparation system ( caprio , 2005 ) .
|
microphysics of weakly magnetized relativistic collisionless shock waves , corroborated by recent high performance numerical simulations , indicate the presence of a microturbulent layer of large magnetic field strength behind the shock front , which must decay beyond some hundreds of skin depths .
the present paper discusses the dynamics of such microturbulence , borrowing from these same numerical simulations , and calculates the synchrotron signature of a powerlaw of shock accelerated particles .
the decaying microturbulent layer is found to leave distinct signatures in the spectro - temporal evolution of the spectrum @xmath0 of a decelerating blast wave , which are potentially visible in early multi - wavelength follow - up observations of gamma - ray bursts .
this paper also discusses the influence of the evolving microturbulence on the acceleration process , with particular emphasis on the maximal energy of synchrotron afterglow photons , which falls in the gev range for standard gamma - ray burst parameters .
finally , this paper argues that the evolving microturbulence plays a key role in shaping the spectra of recently observed gamma - ray bursts with extended gev emission , such as grb090510 .
[ firstpage ] acceleration of particles shock waves gamma - ray bursts
|
* context . * the _ diagonal _ of a multivariate power series with coefficients @xmath3 is the univariate power series with coefficients @xmath4 . particularly interesting is the class of diagonals of _ rational _ power series ( ie , taylor expansions of rational functions ) . in particular , diagonals of _ bivariate _ rational power series are always roots of nonzero bivariate polynomials ( ie , they are algebraic series ) @xcite . since it is also classical that algebraic series are d - finite ( ie , satisfy linear differential equations with polynomial coefficients ) ,
their coefficients satisfy linear recurrences and this leads to an optimal algorithm for the computation of their first terms @xcite . in this article , we determine the degrees of these polynomials , the cost of their computation and related applications
. * previous work . *
the algebraicity of bivariate diagonals is classical .
the same is true for the converse ; also the property persists for multivariate rational series in positive characteristic @xcite .
the first occurrence we are aware of in the literature is plya s article @xcite , which deals with a particular class of bivariate rational functions ; the proof uses elementary complex analysis . along the lines of plya s
approach , furstenberg @xcite gave a ( sketchy ) proof of the general result , over the field of complex numbers ; the same argument has been enhanced later @xcite,@xcite .
three more different proofs exist : a purely algebraic one that works over arbitrary fields of characteristic zero ( * ? ? ?
6.1 ) ( see also ( * ? ? ?
6.3.3 ) ) , one based on non - commutative power series ( * ? ?
5 ) , and a combinatorial proof @xcite . despite the richness of the topic and the fact that most proofs are constructive in essence , we were not able to find in the literature any _ explicit _ algorithm for computing a bivariate polynomial that cancels the diagonal of a general bivariate rational function .
diagonals of rational functions appear naturally in enumerative combinatorics .
in particular , the enumeration of unidimensional walks has been the subject of recent activity , see @xcite and the references therein .
the algebraicity of generating functions attached to such walks is classical as well , and related to that of bivariate diagonals . beyond this structural result
, several quantitative and effective results are known .
explicit formulas give the generating functions in terms of implicit algebraic functions attached to the set of allowed steps in the case of excursions @xcite,@xcite , bridges and meanders @xcite .
moreover , if @xmath5 and @xmath6 denote the upper and lower amplitudes of the allowed steps , the bound @xmath7 on the degrees of equations for excursions has been obtained by bousquet - mlou , and showed to be tight for a specific family of step sets , as well as generically @xcite . from the algorithmic viewpoint , banderier and flajolet gave an algorithm ( called the _ platypus algorithm _ ) for computing a polynomial of degree @xmath8 that annihilates the generating function for excursions @xcite .
* contributions .
* we design ( section [ sec : diagonals ] ) the first explicit algorithm for computing a polynomial equation for the diagonal of an arbitrary bivariate rational function .
we analyze its complexity and the size of its output in theorem [ thm : bound diagonals ] .
the algorithm has two main steps .
the first step is the computation of a polynomial equation for the residues of a bivariate rational function .
we propose an efficient algorithm for this task , that is a polynomial - time version of bronstein s algorithm @xcite ; corresponding size and complexity bounds are given in theorem [ th : bronstein ] .
the second step is the computation of a polynomial equation for the sums of a fixed number of roots of a given polynomial .
we design an additive version of the platypus algorithm @xcite and analyze it in theorem [ thm : platypus - bound ] .
we show in proposition [ prop : generic ] that generically , the size of the minimal polynomial for the diagonal of a rational function is exponential in the degree of the input and that our algorithm computes it in quasi - optimal complexity ( theorem [ thm : bound diagonals ] ) . in the application
to walks , we show how to expand to high precision the generating functions of bridges , excursions and meanders .
our main message is that pre - computing a polynomial equation for them is too costly , since that equation might have exponential size in the maximal amplitude @xmath9 of the allowed steps .
our algorithms have quasi - linear complexity in the precision of the expansion , while keeping the pre - computation step in polynomial complexity in @xmath9 ( theorem [ thm : walks ] ) .
* structure of the paper . * after a preliminary section on background and notation , we first discuss several special bivariate resultants of broader general interest in section [ sec : resultants ] .
next , we consider diagonals , the size of their minimal polynomials and an efficient way of computing annihilating polynomials in section [ sec : diagonals ] .
in this section , that might be skipped at first reading , we introduce notation and technical results that will be used throughout the article . in this article , @xmath10 denotes a field of characteristic 0 .
we denote by @xmath11_n$ ] the set of polynomials in @xmath11 $ ] of degree less than @xmath12 .
similarly , @xmath13 stands for the set of rational functions in @xmath14 with numerator and denominator in @xmath11_n$ ] , and @xmath15_n$ ] for the set of power series in @xmath15 $ ] truncated at precision @xmath12 .
if @xmath16 is a polynomial in @xmath17 $ ] , then its degree with respect to @xmath18 ( resp .
@xmath19 ) is denoted @xmath20 ( resp .
@xmath21 ) , and the _ bidegree _ of @xmath16 is the pair @xmath22 .
the notation @xmath23 is used for univariate polynomials .
inequalities between bidegrees are component - wise .
the set of polynomials in @xmath17 $ ] of bidegree less than @xmath24 is denoted by @xmath17_{n , m}$ ] , and similarly for more variables .
the _ valuation _ of a polynomial @xmath25 $ ] or a power series @xmath26 $ ] is its smallest exponent with nonzero coefficient .
it is denoted @xmath27 , with the convention @xmath28 .
the _ reciprocal _ of a polynomial @xmath29 $ ] is the polynomial @xmath30 .
if @xmath31 , the notation @xmath32 stands for the generating series of the _ newton sums _ of @xmath16 : @xmath33 a _ squarefree decomposition _ of a nonzero polynomial @xmath34 $ ] , where @xmath35 or @xmath11 $ ] , is a factorization @xmath36 , with @xmath37 $ ] squarefree , the @xmath38 s pairwise coprime and @xmath39 . the corresponding _ squarefree part _ of @xmath40 is the polynomial @xmath41 . if @xmath40 is squarefree then @xmath42 .
the coefficient of @xmath43 in a power series @xmath44 $ ] is denoted @xmath45a$ ] . if @xmath46 , then @xmath47 denotes the polynomial @xmath48 .
the exponential series @xmath49 is denoted @xmath50 .
the _ hadamard product _ of two power series @xmath51 and @xmath52 is the power series @xmath53 such that @xmath45a\odot b=[x^n]a\cdot[x^n]b$ ] for all @xmath12 .
if @xmath54 is a bivariate power series in @xmath55 $ ] , the _ diagonal _ of @xmath0 , denoted @xmath1 is the univariate power series in @xmath56 $ ] defined by @xmath57 in several places , we need bounds on degrees of coefficients of bivariate rational series . in most cases ,
these power series belong to @xmath58 $ ] and have a very constrained structure : there exists a polynomial @xmath59 $ ] and an integer @xmath60 such that the power series can be written @xmath61 with @xmath62 $ ] and @xmath63 , for all @xmath12 .
we denote by @xmath64 the set of such power series .
its main properties are summarized as follows .
[ growth ] let @xmath65 $ ] , @xmath66 and @xmath67 $ ] . the set @xmath64 is a subring of @xmath58 $ ] ; let @xmath68 with @xmath69 , then @xmath70 ; the products obey @xmath71 for _ ( 3 ) _ , if @xmath72 and @xmath73 belong respectively to @xmath64 and @xmath74 , then the @xmath12th coefficient of their product is a sum of terms of the form @xmath75 .
therefore , the degree of the numerator is bounded by @xmath76 , whence _ ( 3 ) _ is proved .
property _ ( 1 ) _ is proved similarly . in property _ ( 2 ) _ , the condition on @xmath77 makes @xmath78 well - defined .
the result follows from _ ( 1)_. as consequences , we deduce the following two results .
[ coro : invpol ] let @xmath79 $ ] with @xmath80 be such that @xmath81 .
let @xmath82 be a squarefree part of @xmath40 .
then @xmath83 write @xmath84 with @xmath85 .
then the result when @xmath40 is squarefree ( @xmath86 ) follows from part _ ( 2 ) _ of lemma [ growth ] , with @xmath87 .
the general case then follows from parts _ ( 1,3)_. [ prop : derivatives - rational - function ] let @xmath16 and @xmath40 be polynomials in @xmath17 $ ] , with @xmath88 , @xmath89 and @xmath90 . then for all @xmath91 , @xmath92 with @xmath93 .
the taylor expansion of @xmath94 has for coefficients the derivatives of @xmath0 .
we consider it either in @xmath95 $ ] or in @xmath96 $ ] .
corollary [ coro : invpol ] applies directly for the degree in @xmath18 .
the saving on the degree in @xmath19 follows from observing that in the first part of the proof of the corollary , the decomposition @xmath97 has the property that @xmath98 .
this @xmath99 is then propagated along the proof thanks to part _ ( 3 ) _ of lemma [ growth ] .
we recall classical complexity notation and facts for later use .
let @xmath10 be again a field of characteristic zero .
unless otherwise specified , we estimate the cost of our algorithms by counting arithmetic operations in @xmath10 ( denoted `` ops . '' ) at unit cost .
the soft - o notation @xmath100 indicates that polylogarithmic factors are omitted in the complexity estimates .
we say that an algorithm has quasi - linear complexity if its complexity is @xmath101 , where @xmath9 is the maximal _ arithmetic size _ ( number of coefficients in @xmath10 in a dense representation ) of the input and of the output . in that case , the algorithm is said to be _ quasi - optimal_. * univariate operations .
* throughout this article we will use the fact that most operations on polynomials , rational functions and power series in one variable can be performed in quasi - linear time .
standard references for these questions are the books @xcite and @xcite .
the needed results are summarized in fact [ fact : complexity ] below .
[ fact : complexity ] the following operations can be performed in @xmath102 ops . in @xmath10 : addition , product and differentiation of elements in @xmath11_n$ ] , @xmath13 and @xmath15_n$ ] ; integration in @xmath11_n$ ] and @xmath15_n$ ] ; extended gcd , squarefree decomposition and resultant in @xmath11_n$ ] ; multipoint evaluation in @xmath11_n$ ] , @xmath13 at @xmath103 points in @xmath10 ; interpolation in @xmath11_n$ ] and @xmath13 from @xmath12 ( resp .
@xmath104 ) values at pairwise distinct points in @xmath10 ; inverse , logarithm , exponential in @xmath15_n$ ] ( when defined ) ; conversions between @xmath29_n$ ] and @xmath105_n$ ] . * multivariate operations . *
basic operations on polynomials , rational functions and power series in several variables are hard questions from the algorithmic point of view .
for instance , no general quasi - optimal algorithm is currently known for computing resultants of bivariate polynomials , even though in several important cases such algorithms are available @xcite .
multiplication is the most basic non - trivial operation in this setting .
the following result can be proved using kronecker s substitution ; it is quasi - optimal for fixed number of variables @xmath106 .
[ fact : multiprod ] polynomials in @xmath107_{d_1 , \ldots , d_m}$ ] and power series in @xmath108_{d_1 , \ldots , d_m}$ ] can be multiplied using @xmath109 ops .
a related operation is multipoint evaluation and interpolation .
the simplest case is when the evaluation points form an @xmath110-dimensional tensor product grid @xmath111 , where @xmath112 is a set of cardinal @xmath113 .
[ fact : multieval]@xcite polynomials in @xmath107_{d_1 , \ldots , d_m}$ ] can be evaluated and interpolated from values that they take on @xmath114 points that form an @xmath110-dimensional tensor product grid using @xmath115 ops .
again , the complexity in fact [ fact : multieval ] is quasi - optimal for fixed @xmath106 .
a general ( although non - optimal ) technique to deal with more involved operations on multivariable algebraic objects ( eg , in @xmath17 $ ] ) is to use ( multivariate ) evaluation and interpolation on polynomials and to perform operations on the evaluated algebraic objects using facts [ fact : complexity][fact : multieval ] . to put this strategy in practice , the size of the output needs to be well controlled .
we illustrate this philosophy on the example of resultant computation , based on the following easy variation of ( * ? ? ?
* thm . 6.22 ) .
[ borne resultant ] let @xmath116 and @xmath117 be bivariate polynomials of respective bidegrees @xmath118 and @xmath119 .
then , @xmath120 [ algo resultant ] let @xmath16 and @xmath40 be polynomials in @xmath121_{d_1 , \ldots , d_m , d}$ ]
. then @xmath122 belongs to @xmath107_{d_1 , \ldots , d_m}$ ] , where @xmath123 .
moreover , the coefficients of @xmath124 can be computed using @xmath125 ops . in @xmath10 .
the degrees estimates follow from fact [ borne resultant ] . to compute @xmath124
, we use an evaluation - interpolation scheme : @xmath16 and @xmath40 are evaluated at @xmath126 points @xmath127 forming an @xmath110 dimensional tensor product grid ; @xmath128 univariate resultants in @xmath129_d$ ] are computed ; @xmath124 is recovered by interpolation .
by fact [ fact : multieval ] , the evaluation and interpolation steps are performed in @xmath130 ops .
the second one has cost @xmath131 .
using the inequality @xmath132 concludes the proof . .
we conclude this section by recalling a complexity result for the computation of a squarefree decomposition of a bivariate polynomial .
[ fact : sqfree]@xcite a squarefree decomposition of a polynomial in @xmath17_{d_x , d_y}$ ] can be computed using @xmath133 ops .
we are interested in a polynomial that vanishes at the residues of a given rational function .
it is a classical result in symbolic integration that in the case of simple poles , there is a resultant formula for such a polynomial , first introduced by rothstein @xcite and trager @xcite .
this was later generalized by bronstein @xcite to accommodate multiple poles as well .
however , as mentioned by bronstein , the complexity of his method grows exponentially with the multiplicity of the poles .
instead , we develop in this section an algorithm with polynomial complexity .
let @xmath134 be a nonzero element in @xmath135 , where @xmath136 are two coprime polynomials in @xmath129 $ ] .
let @xmath137 be a squarefree decomposition of @xmath40 .
for @xmath138 , if @xmath139 is a root of @xmath38 in an algebraic extension of @xmath10 , then it is simple and the residue of @xmath140 at @xmath139 is the coefficient of @xmath141 in the laurent expansion of @xmath142 at @xmath143 .
if @xmath144 is the polynomial @xmath145 , this residue is the coefficient of @xmath146 in the taylor expansion at @xmath143 of the regular rational function @xmath147 , computed with rational operations only and then evaluated at @xmath148 . if this coefficient is denoted @xmath149 , with polynomials @xmath150 and @xmath151 , the residue at @xmath139 is a root of @xmath152 . when @xmath153 , this is exactly the rothstein - trager resultant
this computation leads to algorithm [ algo : bronstein ] , which avoids the exponential blowup of the complexity that would follow from a symbolic pre - computation of the bronstein resultants .
algorithm * * algebraicresidues**@xmath154 two polynomials @xmath16 and @xmath155$]a polynomial in @xmath156 $ ] canceling all the residues of @xmath157 compute @xmath158 a squarefree decomposition of @xmath40 ; @xmath159 @xmath160 ; @xmath161 ; expand @xmath162 ; write @xmath163 as @xmath164 with @xmath150 and @xmath151 coprime ; @xmath165 ; @xmath166
[ ex : bronstein ] let @xmath167 be an integer , and let @xmath168 $ ] be the rational function @xmath169 .
the poles have order @xmath170 . in this example , the algorithm can be performed by hand for arbitrary @xmath9 : a squarefree decomposition has @xmath171 and @xmath172 , the other @xmath38 s being 1 . then @xmath173 and the next step is to expand @xmath174 expanding the binomial series gives the coefficient of @xmath175 as @xmath176 , with @xmath177 the residues
are then cancelled by @xmath178 , namely @xmath179 * bounds . * in our applications , as in the previous example , the polynomials @xmath16 and @xmath40 have coefficients that are themselves polynomials in another variable @xmath18 .
let then @xmath180 , @xmath181 , @xmath182 and @xmath183 be the bidegrees in @xmath184 of @xmath16 , @xmath40 , @xmath82 and @xmath38 , where @xmath41 is a squarefree part of @xmath40 . in algorithm
[ algo : bronstein ] , @xmath185 has degree at most @xmath186 in @xmath18 and total degree @xmath187 in @xmath188 .
similarly , @xmath189 has degree @xmath190 in @xmath18 and total degree @xmath191 in @xmath188 . when @xmath192 , by proposition [ prop : derivatives - rational - function ] , the coefficient @xmath193 in the power series expansion of @xmath194 has denominator of bidegree bounded by @xmath195 and numerator of bidegree bounded by @xmath196 .
thus by fact [ borne resultant ] , @xmath197 is at most @xmath198 while its degree in @xmath199 is bounded by the number of residues @xmath200 .
summing over all @xmath201 leads to the bound @xmath202 if @xmath203 , a direct computation gives the bound @xmath204 .
[ th : bronstein ] let @xmath205 .
let @xmath82 be a squarefree part of @xmath40 wrt y. let @xmath206 be bounds on the bidegree of @xmath82 . then the polynomial computed by algorithm [ algo : bronstein ] annihilates the residues of @xmath157 , has degree in @xmath199 bounded by @xmath207 and degree in @xmath18 bounded by @xmath208 it can be computed in @xmath209 operations in @xmath10 .
note that both bounds above ( when @xmath192 and @xmath203 ) are upper bounded by @xmath210 , independently of the multiplicities .
the complexity is also bounded independently of the multiplicities by @xmath211 .
the bounds on the bidegree of @xmath212 are easily derived from the previous discussion .
by fact [ fact : sqfree ] , a squarefree decomposition of @xmath40 can be computed using @xmath133 ops .
we now focus on the computations performed inside the @xmath201th iteration of the loop .
computing @xmath213 requires an exact division of polynomials of bidegrees at most @xmath214 ; this division can be performed by evaluation - interpolation in @xmath215 ops .
similarly , the trivariate polynomial @xmath185 can be computed by evaluation - interpolation wrt @xmath184 in time @xmath216 . by the discussion preceding theorem [ th : bronstein ] , both @xmath217 and @xmath218 have bidegrees at most @xmath219 , where @xmath220 and @xmath221 .
they can be computed by evaluation - interpolation in @xmath222 ops .
finally , the resultant @xmath223 has bidegree at most @xmath224 , and since the degree in @xmath19 of @xmath225 and @xmath38 is at most @xmath226 , it can be computed by evaluation - interpolation in @xmath227 ops by lemma [ algo resultant ] .
the total cost of the loop is thus @xmath228 , where @xmath229 using the ( crude ) bounds @xmath230 , @xmath231 , @xmath232 and @xmath233 shows that @xmath234 is bounded by @xmath235 which , by using the inequalities @xmath236 and @xmath237 , is seen to belong to @xmath238 .
gathering together the various complexity bounds yields the stated bound and finishes the proof of the theorem . * remark . *
note that one could also use hermite reduction combined with the usual rothstein - trager resultant in order to compute a polynomial @xmath239 that annihilates the residues .
indeed , hermite reduction computes an auxiliary rational function that admits the same residues as the input , while only having simple poles .
a close inspection of this approach provides the same bound @xmath207 for the degree in @xmath19 of @xmath239 , but a less tight bound for its degree in @xmath18 , namely worse by a factor of @xmath207 .
the complexity of this alternative approach appears to be @xmath240 ( using results from @xcite ) , to be compared with the complexity bound from theorem [ th : bronstein ] .
given a polynomial @xmath241 $ ] of degree @xmath9 with coefficients in a field @xmath10 of characteristic 0 , let @xmath242 be its roots in the algebraic closure of @xmath10 . for any positive integer @xmath243 ,
the polynomial of degree @xmath244 defined by @xmath245 has coefficients in @xmath10 .
this section discusses the computation of @xmath246 summarized in algorithm [ algo : sigma_c ] , which can be seen as an additive analogue of the _ platypus algorithm _ of banderier and flajolet @xcite .
algorithm * * purecomposedsum**@xmath247 a polynomial @xmath16 of degree @xmath9 in @xmath129 $ ] , a positive integer @xmath248the polynomial @xmath249 from eq .
@xmath250 @xmath251 @xmath252 @xmath253 @xmath254f)\odot\sum{n!y^n}\bmod y^{d+1}$ ] @xmath255 we recall two classical formulas ( see , eg , @xcite ) , the second one being valid for monic @xmath16 only : : @xmath256 truncating these formulas at order @xmath170 makes @xmath32 a representation of the polynomial @xmath16 ( up to normalization ) , since both conversions above can be performed quasi - optimally by newton iteration @xcite .
the key for algorithm [ algo : sigma_c ] is the following variant of @xcite .
let @xmath241 $ ] be a polynomial of degree @xmath9 , let @xmath32 denote the generating series of its newton sums and let @xmath257 be the series @xmath258 .
let @xmath259 be the polynomial in @xmath260 $ ] defined by @xmath261\exp\left(\sum_{n \ge 1}{(-1)^{n-1}t_n\frac{z^n}{n } } \right).\ ] ] then the following equality holds @xmath262 by construction , the series @xmath257 is @xmath263 when applied to the polynomial @xmath249 , this becomes @xmath264\prod_{i=1}^{d}{\left(1+z \exp(\alpha_i y)\right)}. \end{aligned}\ ] ] this expression rewrites : @xmath265\exp\left(\sum_{i=1}^{d}\log(1+z \exp({\alpha_i y}))\right)\\ = [ z^c]\exp\left(\sum_{i=1}^{d}\sum_{m\ge 1}{(-1)^{m-1}\exp({\alpha_i m y})\frac{z^m}{m}}\right)\\ = [ z^c]\exp\left(\sum_{m\ge 1}{(-1)^{m-1}s(m y)\frac{z^m}{m}}\right ) , \end{gathered}\ ] ] and the last expression equals @xmath266 .
the correctness of algorithm [ algo : sigma_c ] follows from observing that the truncation orders @xmath267 in @xmath19 and @xmath268 in @xmath199 of the power series involved in the algorithm are sufficient to enable the reconstruction of @xmath246 from its first newton sums by .
* bivariate case .
* we now consider the case where @xmath16 is a polynomial in @xmath17 $ ] .
then , the coefficients of @xmath246 wrt @xmath19 may have denominators .
we follow the steps of algorithm [ algo : sigma_c ] ( run on @xmath16 viewed as a polynomial in @xmath19 with coefficients in @xmath14 ) in order to compute bounds on the bidegree of the polynomial obtained by clearing out these denominators .
we obtain the following result .
[ thm : platypus - bound ] let @xmath269_{d_x+1,d_y+1}$ ] , let @xmath270 be a positive integer such that @xmath271 and let @xmath272 .
let @xmath273 $ ] denote the leading coefficient of @xmath16 wrt @xmath19 and let @xmath249 be defined as in eq . .
then @xmath274 is a polynomial in @xmath17 $ ] of bidegree at most @xmath275 that cancels all sums @xmath276 of @xmath270 roots @xmath277 of @xmath16 , with @xmath278 .
moreover , this polynomial can be computed in @xmath279 ops .
this result is close to optimal .
experiments suggest that for generic @xmath16 of bidegree @xmath214 the minimal polynomial of @xmath276 has bidegree @xmath280 .
in particular , our degree bound is precise in @xmath19 , and overshoots by a factor of @xmath281 only in @xmath18 .
similarly , the complexity result is quasi - optimal up to a factor of @xmath282 only .
the newton series @xmath283 has the form @xmath284 with @xmath285 .
since both factors belong to @xmath286 , lemma [ growth ] implies that @xmath287 . applying this same lemma repeatedly
, we get that @xmath288 ( stability under the integration of algorithm [ algo : sigma_c ] is immediate ) .
since @xmath249 has degree @xmath128 wrt @xmath19 , we deduce that @xmath289 is a polynomial that satisfies the desired bound . by evaluation and interpolation at @xmath290 points , and newton iteration for quotients of power series in @xmath291_{1+d}$ ] ( fact [ fact : complexity ] ) , the power series @xmath283 can be computed in @xmath292 ops . the power series @xmath257
is then computed from @xmath283 in @xmath293 ops . to compute @xmath0 we use evaluation - interpolation wrt @xmath18 at @xmath290 points , and fast exponentials of power series ( fact [ fact : complexity ] ) .
the cost of this step is @xmath279 ops .
then , @xmath294 is computed for @xmath293 additional ops .
the last exponential is again computed by evaluation - interpolation and newton iteration using @xmath292 ops .
the relation between diagonals of bivariate rational functions and algebraic series is classical @xcite .
we recall here the usual derivation when @xmath295 while setting our notation .
let @xmath296 be a rational function in @xmath297 , whose denominator does not vanish at @xmath298
. then the diagonal of @xmath0 is a convergent power series that can be represented for small enough @xmath299 by a cauchy integral @xmath300 where the contour is for instance a circle of radius @xmath301 inside an annulus where @xmath302 remains in the domain of convergence of @xmath0 .
this is the basis of an algebraic approach to the computation of the diagonal as a sum of residues of the rational function @xmath303 with @xmath16 and @xmath40 two coprime polynomials . for @xmath299 small enough
, the circle can be shrunk around 0 and only the roots of @xmath304 tending to 0 when @xmath305 lie inside the contour @xcite .
these are called the _
small branches_. thus the diagonal is given as @xmath306 where the sum is over the _ distinct _ roots of @xmath40 tending to 0 .
we call their number the number of small branches of @xmath40 and denote it by @xmath307
. since the @xmath308 s are algebraic and finite in number and residues are obtained by series expansion , which entails only rational operations , it follows that the diagonal is algebraic too .
combining the algorithms of the previous section gives algorithm [ algo : diagonal ] that produces a polynomial equation for @xmath1 .
the correctness of this algorithm over an arbitrary field of characteristic 0 follows from an adaptation of the arguments of gessel and stanley ( * ? ? ?
6.1 ) , ( * ? ?
6.3.3 ) .
let @xmath167 be an integer , and let @xmath309 be the rational function @xmath310 .
the diagonal of @xmath311 is equal to @xmath312 by the previous argument , it is an algebraic series , which is the sum of the residues of the rational function @xmath313 of example [ ex : bronstein ] over its small branches ( with @xmath18 replaced by @xmath299 ) . in this case , the denominator is @xmath314 .
it has one solution tending to 0 with @xmath299 ; the other one tends to @xmath315 .
thus the diagonal is cancelled by the quadratic polynomial .
algorithm * algebraicdiagonal*(@xmath316 ) two polynomials @xmath51 and @xmath317 $ ] , with @xmath318a polynomial @xmath319 $ ] such that @xmath320 @xmath321 write @xmath322 as @xmath157 with coprime polynomials @xmath16 and @xmath40 ; @xmath323 @xmath324 number of small branches of @xmath40 @xmath325 @xmath326 for an integer @xmath327 , we consider the rational function @xmath328 of bidegree @xmath329 .
the first step of the algorithm produces @xmath330 whose denominator is irreducible with @xmath9 small branches . running algorithm [ algo : diagonal ] on this example
, we obtain a polynomial @xmath331 annihilating @xmath332 , which is experimentally irreducible and whose bidegrees for @xmath333 are @xmath334 . from these values , it is easy to conjecture that the bidegree is given by @xmath335 of exponential growth in the bidegree of @xmath311 . in general , these bidegrees do not grow faster than in this example . in theorem [ thm : bound diagonals ] , we prove bounds that are barely larger than the values above .
the rest of this section is devoted to the derivation of bounds on the complexity of algorithm [ algo : diagonal ] and on the size of the polynomial it computes , which are given in theorem [ thm : bound diagonals ] .
* degrees . *
a bound on the bidegree of @xmath336 will be obtained from the bounds successively given by theorems [ th : bronstein ] and [ thm : platypus - bound ] . in order to follow the impact of the change of variables in the first step , we define the _ diagonal degree _ of a polynomial @xmath337 as the integer
@xmath338 we collect the properties of interest in the following .
[ lemma : slope ] for any @xmath16 and @xmath40 in @xmath17 $ ] , @xmath339 ; @xmath340 ; there exists a polynomial @xmath341 $ ] , such that + @xmath342 , with @xmath343 and @xmath344 @xmath345 .
part _ ( 1 ) _ is immediate .
the quantity @xmath346 is nothing else than @xmath347 , which makes parts _ ( 2 ) _ and _ ( 3 ) _ clear too .
from there , we get the identity @xmath348 for arbitrary @xmath16 and @xmath40 , whence @xmath349 and part _ ( 4 ) _ is a consequence of parts _ ( 1 ) _ and _ ( 3)_. thus , starting with a rational function @xmath350 , with @xmath214 a bound on the bidegrees of @xmath51 and @xmath52 , and @xmath206 a bound on the bidegree of a squarefree part @xmath351 of @xmath52 , the first step of the algorithm constructs @xmath352 , with polynomials @xmath16 and @xmath40 and @xmath353 these inequalities give bounds on the degrees in @xmath18 of the numerator and denominator of @xmath322 .
the rest of the computation depends on the sign of @xmath139 .
if @xmath354 , then the degrees in @xmath19 of @xmath355 and @xmath40 are bounded by @xmath356 , while if @xmath357 , those of @xmath16 and @xmath358 are bounded by @xmath359 .
thus in both cases they are bounded by @xmath360 , where @xmath361 a squarefree part of the denominator has degree in @xmath19 bounded by @xmath362 .
from there , theorem [ th : bronstein ] yields @xmath363 , with @xmath364 * small branches .
* it is classical that for a polynomial @xmath365 $ ] , the number of its solutions tending to 0 can be read off its newton polygon .
this polygon is the lower convex hull of the union of @xmath366 for @xmath367 such that @xmath368 .
the number of solutions tending to 0 is given by the minimal @xmath19-coordinate of its leftmost points .
since the number of small branches counts only distinct solutions , it is thus given by @xmath369p^\star).\ ] ] the change of variables @xmath370 changes the coordinates of the point corresponding to @xmath371 into @xmath372 .
this transformation maps the vertices of the original newton polygon to the vertices of the newton polygon of the laurent polynomial @xmath373 . multiplying by @xmath374 yields a polynomial and shifts the newton polygon up by @xmath346 , thus @xmath375 the number of small branches of the denominator of @xmath322 constructed in the first step of the algorithm
is then given by @xmath376 * complexity .
* we now analyze the cost of algorithm [ algo : diagonal ] .
the first step does not require any arithmetic operation .
next , the computation of @xmath124 takes @xmath377 ops .
( see the comment after theorem [ th : bronstein ] ) .
the number of small branches is obtained with no arithmetic operation from a squarefree decomposition computed in algorithm [ algo : bronstein ] .
finally , algorithm [ algo : sigma_c ] uses @xmath378 ops .
we now have the values required by theorem [ thm : platypus - bound ] , which concludes the proof of the following bounds .
[ thm : bound diagonals ] let @xmath379 be a rational function in @xmath380 with @xmath318 .
let @xmath214 ( resp .
@xmath206 ) be a bound on the bidegrees of @xmath51 and @xmath52 ( resp . a squarefree part of @xmath52 ) .
let @xmath381 be defined as in eqs .
( [ eq : epsilon],[eq : dxdy],[eq : nb_small ] ) . then there exists a polynomial @xmath319 $ ] such that @xmath382 and @xmath383 algorithm [ algo : diagonal ] computes it in @xmath384 ops .
a general bound on @xmath385 depending only on a bound @xmath386 on the bidegree of the input can be deduced from the above as @xmath387 assume that the denominator of @xmath388 is already partially factored as @xmath389 , where the @xmath308 are @xmath390 distinct _ rational _ branches among the @xmath270 small branches of @xmath40 .
then their corresponding ( rational ) residues @xmath391 contribute to the diagonal ; therefore it is only necessary to invoke algorithm [ algo : diagonal ] on @xmath392 , which produces a polynomial @xmath393 .
then the polynomial @xmath394 cancels the diagonal of @xmath0 .
in particular , this optimization applies systematically for the factor @xmath395 when @xmath357 ( or equivalently @xmath396 ) in the algorithm . in this case
, it yields a polynomial @xmath336 with smaller degree than the original algorithm : @xmath397 ( a sharper bound on the degree in @xmath299 can be derived as well . ) the bounds from theorem [ thm : bound diagonals ] on the bidegree of @xmath336 are slightly pessimistic wrt the variable @xmath299 , but generically tight wrt the variable @xmath398 , as will be proved in proposition [ prop : generic ] below .
we first need a lemma .
[ lemma : galois groups ] let @xmath10 be a field of characteristic @xmath399 , and @xmath400 $ ] be a polynomial of degree @xmath9 , with galois group @xmath401 over @xmath10 .
assume that the roots @xmath402 of @xmath16 are algebraically independent over @xmath403 .
then , for any @xmath248 , the degree @xmath244 polynomial @xmath246 is irreducible in @xmath129 $ ] .
since @xmath404 is a root of @xmath249 , it suffices to prove that @xmath405 has degree @xmath244 over @xmath10 .
the @xmath406 s being algebraically independent , any permutation @xmath407 of all the @xmath406 s that leaves @xmath408 unchanged has to preserve @xmath409 as well .
it follows that @xmath410 has degree @xmath411 over @xmath405 and degree @xmath412 over @xmath10 , so that @xmath405 has degree @xmath244 over @xmath10 [ prop : generic ] let @xmath51 be a polynomial in @xmath413_{d_x , d_y}$ ] , and @xmath414,\ ] ] where the @xmath415 are indeterminates .
then the polynomial computed by algorithm [ algo : diagonal ] with input @xmath316 is irreducible of degree @xmath416 over @xmath417 .
first apply the change of variables to obtain @xmath418 , with @xmath419 .
denote @xmath420 .
then , the polynomial @xmath421 has the form @xmath422 where the @xmath423 s are algebraically independent over @xmath403 .
therefore , @xmath421 has galois group @xmath424 over @xmath425 and its roots are algebraically independent over @xmath403 @xcite .
this property lifts to @xmath117 @xcite , which thus has galois group @xmath401 and algebraically independent roots , denoted @xmath426 .
now define the polynomial @xmath427 . since @xmath40 has simple roots , this is exactly the polynomial that is computed by algorithm [ algo : bronstein ] .
the family @xmath428 is algebraically independent , since any algebraic relation between them would induce one for the @xmath308 s by clearing out denominators . in particular ,
the natural morphism @xmath429 is injective , whence an isomorphism .
( here , @xmath430 denotes the galois group of @xmath241 $ ] over @xmath10 . ) since an immediate investigation of the newton polygon of @xmath40 shows that it has @xmath431 small branches , we conclude using lemma [ lemma : galois groups ] .
proposition [ prop : generic ] implies that for a generic rational function @xmath316 with @xmath432_{d , d}$ ] and @xmath317_{d+1,d+1}$ ] , the degree of @xmath336 in @xmath398 is @xmath433 .
this is indeed observed on random examples .
we consider a rational function @xmath434 , where @xmath435 is a dense polynomial of bidegree @xmath386 chosen at random .
for @xmath333 , algorithm * algebraicdiagonal*(@xmath0 ) produces _ irreducible _ outputs with bidegrees @xmath436 , @xmath437 , @xmath438 , @xmath439 , that are matched by the formulas
@xmath440 so that the bound on @xmath441 is tight in this case and the irreducibility of the output shows that theorem [ thm : bound diagonals ] can not be improved further .
the exponential degree of the minimal polynomial of a diagonal proved in proposition [ prop : generic ] concerns more generally other sums of residues , since this is the step where the exponential growth of the algebraic equations appears .
this includes in particular constant terms of rational functions in @xmath442 $ ] , that can also be written as contour integrals of rational functions around the origin .
by contrast , sums of residues of a rational function always satisfy a differential equation of only polynomial size @xcite .
thus , when an algebraic function appears to be connected to a sum of residues of a rational function , the use of this differential structure is much more adapted to the computation of series expansions , instead of going through a potentially large polynomial . as an example where this phenomenon occurs naturally , we consider here the enumeration of unidimensional lattice walks , following banderier and flajolet @xcite and bousquet - mlou @xcite .
our goal in this section is to study , from the algorithmic perspective , the series expansions of various generating functions ( for bridges , excursions , meanders ) that have been identified as algebraic @xcite .
one of our contributions is to point out that although algebraic series can be expanded fast @xcite , the pre - computation of a polynomial equation could have prohibitive cost .
we overcome this difficulty by pre - computing differential ( instead of polynomial ) equations that have polynomial size only , and using them to compute series expansions to precision @xmath2 for bridges , excursions and meanders in time quasi - linear in @xmath2 .
we start with some vocabulary on lattice walks .
a _ simple step _ is a vector @xmath443 with @xmath444 .
@xmath257 is a finite set of simple steps .
a _ unidimensional walk _ in the plane @xmath445 built from @xmath257 is a finite sequence @xmath446 of points in @xmath445 , such that @xmath447 and @xmath448 with @xmath449 . in this case
@xmath12 is called the _ length _ of the walk , and @xmath257 is the _ step set _ of the walk .
the @xmath19-coordinate of the endpoint @xmath450 , namely @xmath451 , is called the final altitude of the walk .
the characteristic polynomial of the step set @xmath257 is @xmath452 following banderier and flajolet , we consider three specific families of walks : bridges , excursions and meanders @xcite .
_ bridges _ are walks with final altitude @xmath399 , _ meanders _ are walks confined to the upper half plane , and _ excursions _ are bridges that are also meanders .
we define the full generating power series of walks @xmath453[[x]],\ ] ] where @xmath454 is the number of walks with step set @xmath257 , of length @xmath12 and final altitude @xmath390 .
we denote by @xmath455 ( resp .
@xmath456 , and @xmath457 ) the power series @xmath458 , where @xmath459 is the number of bridges ( resp .
excursions , and meanders ) of length @xmath12 with step set @xmath257 .
our main objective in what follows is to study the efficiency of computing the power series expansions of the series @xmath52 , @xmath461 and @xmath462 . in the next two sections , we first study two previously known methods
, then we design a new one .
we denote by @xmath470 ( resp .
@xmath471 ) the largest @xmath472 such that @xmath473 ( resp .
@xmath474 ) and denote by @xmath9 the sum @xmath475 .
the integer @xmath9 measures the vertical amplitude of @xmath257 ; this makes @xmath9 a good scale for measuring the complexity of the algorithms that will follow .
we assume that both @xmath470 and @xmath471 are positive , since otherwise the study of the excursions and meanders becomes trivial . * the direct method . *
the combinatorial definition of walks yields a recurrence relation for @xmath454 : @xmath476 with initial conditions @xmath477 if @xmath478 with @xmath479 , and @xmath480 .
if @xmath481 denotes the number of walks of length @xmath12 and final altitude @xmath390 that never exit the upper half plane , then @xmath481 also satisfies recurrence , but with the additional initial conditions @xmath482 for all @xmath483 .
then the bridges ( resp . excursions , meanders ) are counted by the numbers @xmath484 ( resp .
@xmath485 , @xmath486 ) .
one can compute these numbers by unrolling the recurrence relation .
each use of the recurrence costs @xmath487 ops . , and in the worst case one has to compute @xmath488 terms of the sequence ( for example , if the step set is @xmath489 ) .
this leads to the computation of each of the generating series in @xmath490 ops . * using algebraic equations . *
another method is suggested in @xcite .
it relies on the algebraicity of @xmath52 , @xmath461 and @xmath462 ( fact [ fact : walks]_(2 ) _ ) .
the series @xmath461 and @xmath462 can be expressed as products in terms of the small branches of the characteristic polynomial @xmath491 ( see ( * ? ? ?
1 ) ) . from there
, a polynomial equation can be obtained using the platypus algorithm @xcite , which computes a polynomial canceling the products of a fixed number of roots of a given polynomial . given a polynomial equation @xmath492 ,
another one for @xmath52 can be deduced from the relation @xmath493 as @xmath494 .
once a polynomial equation is known for one of these three series , it can be used to compute a linear recurrence with polynomial coefficients satisfied by its coefficients @xcite .
this method produces an algorithm that computes the first @xmath2 terms of @xmath52 , @xmath461 and @xmath462 in @xmath495 ops . for this to be an improvement over the naive method for large @xmath2
, the dependence on @xmath9 of the constant in the @xmath496 should not be too large and the precomputation not too costly . indeed , the cost of the pre - computation of an algebraic equation is not negligible .
generically , the minimal polynomial of @xmath461 has degree @xmath497 , which may be exponentially large with respect to @xmath9 @xcite .
empirically , the polynomials for @xmath52 and @xmath462 are similarly large .
the situation for differential equations and recurrences is different : @xmath52 satisfies a differential equation of only polynomial size ( see below ) , whereas ( empirically ) , those for @xmath461 and @xmath462 have a potentially exponential size .
these sizes then transfer to the corresponding recurrences and thereby to the constant in the complexity of unrolling them . with the step set @xmath498 and @xmath499
, the counting series @xmath500 equals @xmath501 experiments indicate that the minimal polynomial of @xmath455 has bidegree @xmath502 , exhibiting an exponential growth in @xmath9 . on the other hand , they show that @xmath455 satisfies a linear differential equation of order @xmath503 and coefficients of degree @xmath504 for even @xmath9 , and @xmath505 for odd @xmath9 . * new method .
* we now give a method that runs in quasi - linear time ( with respect to @xmath2 ) and avoids the computation of an algebraic equation .
our method relies on the fact that periods of rational functions such as the one in part _ ( 3 ) _ of fact [ fact : walks ] satisfy differential equations of polynomial size in the degree of the input rational function @xcite .
we summarize our results in the following theorem , and then go over the proof in each case individually .
[ thm : walks ] let @xmath257 be a finite set of simple steps and @xmath506 .
the series @xmath507 ( resp . @xmath508 and @xmath509 ) can be expanded at order @xmath2 in @xmath510 ops .
( resp . @xmath511 ops . ) , after a pre - computation in @xmath512 ops .
* bridges .
* to expand @xmath465 , we rely on fact [ fact : walks]_(3)_. the formula can be written @xmath513 , the integration path being a circle inside a small annulus around the origin ( * ? ? ?
* proof of th . 1 ) .
moreover , @xmath514 is of the form @xmath157 , where @xmath515 and @xmath516 .
since @xmath16 and @xmath40 are relatively prime and @xmath40 is primitive with respect to @xmath19 , algorithm * hermitetelescoping * ( * ? ? ?
3 ) computes a telescoper for @xmath157 , which is also a differential equation satisfied by @xmath52 , in @xmath512 ops .
the resulting differential equation has order at most @xmath9 and degree @xmath517 .
this differential equation can be turned into a recurrence of order @xmath517 in quasi - optimal time ( see the discussion after ( * ? ? ?
we may use it to expand @xmath518 in @xmath519 ops , once we have a way to compute the initial conditions .
but this can be done using the naive algorithm described above in @xmath520 ops .
thus , the total cost of the pre - computation is @xmath512 , as announced . a set @xmath257 of simple steps and an integer @xmath2@xmath521 @xmath522 [ case @xmath523 or @xmath524 [ case @xmath462 ] @xmath526 ( * ? ?
3 ) @xmath527 the recurrence of order @xmath301 associated to @xmath128 @xmath528w(x , y)\bmod x^{r+1}$ ] [ case @xmath529 + @xmath530yw(x , y)/(1-y)\bmod x^{r+1}$ ] [ case @xmath462 ] @xmath531w(x , y)\bmod x^{n+1}$ ] ( from @xmath532 ) @xmath533yw(x , y)/(1-y)\bmod x^{n+1}$ ] ( from @xmath532 ) @xmath534 @xmath535 @xmath536 * excursions . * if @xmath537 is known , it is then possible to recover @xmath538 thanks to fact [ fact : walks]_(4)_. expanding @xmath466 comes down to the computation of the exponential of a series , which can be performed using @xmath539 ops .
( fact [ fact : complexity]_(4 ) _ ) . denote by @xmath541 the small branches of the polynomial @xmath542 .
then @xmath462 is given as ( * ? ? ?
1 ) : @xmath543 on the other hand , @xmath544 where the integral has been taken over a circle around the origin and the small branches . differentiating the equation @xmath545 = 0 with respect to @xmath18 leads to @xmath546 , whence @xmath547 therefore , @xmath548 , finishing the proof .
thus we apply the same method as in the case of the excursions .
we first compute a differential equation for @xmath549 using the method of @xcite .
the computation of the initial conditions for @xmath51 can also be performed naively from its definition as a constant term , by simply expanding @xmath550 .
the formula of the proposition then recovers @xmath467 .
the complexity analysis goes exactly as in the previous case , giving a global cost of @xmath512 ops .
|
the diagonal of a multivariate power series @xmath0 is the univariate power series @xmath1 generated by the diagonal terms of @xmath0 .
diagonals form an important class of power series ; they occur frequently in number theory , theoretical physics and enumerative combinatorics .
we study algorithmic questions related to diagonals in the case where @xmath0 is the taylor expansion of a bivariate rational function .
it is classical that in this case @xmath1 is an algebraic function .
we propose an algorithm that computes an annihilating polynomial for @xmath1 .
generically , it is its minimal polynomial and is obtained in time quasi - linear in its size .
we show that this minimal polynomial has an exponential size with respect to the degree of the input rational function .
we then address the related problem of enumerating directed lattice walks .
the insight given by our study leads to a new method for expanding the generating power series of bridges , excursions and meanders .
we show that their first @xmath2 terms can be computed in quasi - linear complexity in @xmath2 , without first computing a very large polynomial equation .
= 10000 = 10000 * categories and subject descriptors : * + i.1.2 [ * computing methodologies * ] : symbolic and algebraic manipulations _ algebraic algorithms _ * general terms : * algorithms , theory .
* keywords : * diagonals , walks , algorithms .
|
crum s seminal paper of 1955 @xcite has played an essential role in elucidating the structure of one - dimensional quantum mechanical systems in general and exactly solvable ones , in particular . throughout this paper ,
we mean ` exact solvability ' in the schrdinger picture , namely a quantum system is exactly solved when the complete set of eigenvalues and eigenfunctions are known .
many exactly solvable quantum mechanical hamiltonians were constructed and investigated by combining shape invariance @xcite and crum s theorem @xcite , or the factorisation method @xcite or the method of the so - called supersymmetric quantum mechanics @xcite .
it is interesting to note that most of these shape invariant systems are also solvable in the heisenberg picture @xcite .
exactly solvable quantum mechanical systems of one and many degrees of freedom are not only important in their own right but also have fundamental applications in various disciplines of physics / mathematics , _
e.g. _ the fokker - planck equations @xcite and their discretised version , birth and death processes @xcite , to name a few .
shape invariance is a sufficient condition for exactly solvable quantum mechanical systems .
the number of shape invariant systems , however , was quite limited ; only about a dozen until the recent discovery @xcite of the several types of infinitely many shape invariant hamiltonians @xcite which led to the infinitely many exceptional laguerre , jacobi , wilson and askey - wilson polynomials .
many methods were proposed to derive exactly solvable but non - shape invariant quantum mechanical systems from known shape invariant ones @xcite .
( we apologise to those whose work we have missed . ) among them krein - adler s modification @xcite of crum s theorem is the most comprehensive way to generate infinitely many variants of exactly solvable hamiltonians and their eigenfunctions , starting from an exactly solvable one . the derived system is _ iso - spectral _ with the original one except that a finite number of energy levels are deleted
. if the original system has polynomial eigenfunctions , as is usually the case , the derived systems have also polynomial eigenfunctions . by construction , these polynomials constitute a complete set of orthogonal functions .
but they do not qualify to be called _ exceptional orthogonal polynomials _
@xcite since some members of certain degrees are _ missing _ due to the _
deletion_. the discrete quantum mechanics is a deformation of the ordinary quantum mechanics in the sense that the schrdinger equation is a second order _ difference _ equation instead of differential . in the formulation of odake and sasaki @xcite , the algebraic and analytical structure of quantum mechanics as well as shape invariance and exact solvability
are retained in the discrete version .
the eigenfunctions of the exactly solvable one - dimensional discrete quantum mechanics are the askey - scheme of hypergeometric orthogonal polynomials and their @xmath2-versions @xcite , _
e.g. _ the continuous hahn , the wilson and the askey - wilson polynomials .
these examples are all shape invariant and they are also solvable in the heisenberg picture @xcite .
the dynamical symmetry algebra of these algebras are the askey - wilson algebras @xcite and degenerations , which contain the @xmath2-oscillator algebra @xcite .
the discrete version of crum s theorem was also established recently @xcite . in this paper
we present the discrete quantum mechanics version of adler s modification @xcite of crum s theorem .
it allows to generate an infinite variety of exactly solvable discrete quantum hamiltonian systems . the insight obtained from crum s theorems and their modification , in
the ordinary and the discrete quantum mechanics , is essential for the recent derivation of the infinite numbers of shape invariant systems and the new exceptional orthogonal polynomials @xcite
. we will discuss the main results , the specialisation to the cases of polynomial eigenfunctions and simplest example for various exactly solvable cases ; first for the ordinary quantum mechanics and then for the discrete versions .
the reason is two - fold ; firstly to introduce appropriate notion and notation in the familiar cases of the ordinary quantum mechanics . secondly we choose to reveal the underlying logical processes which are not easy to fathom in adler s paper @xcite or in crum s original article @xcite .
as seen in the subsequent sections , the logical structures of the associated hamiltonian systems and their modification by _ deletion _ of energy levels are shared by the ordinary and the discrete quantum mechanics .
this paper is organised as follows . in section two ,
adler s modification of crum s theorem is recapitulated in appropriate notation for our purposes .
the specialisation to the cases of polynomial eigenfunctions is discussed in some detail .
section three provides the discrete quantum mechanics version of the modification of crum s theorem . again
the specialisation to the cases of polynomial eigenfunctions is mentioned .
appendix gives the simplest examples of the modified hamiltonian systems obtained by deleting the lowest lying @xmath3 excited states for various exactly solvable hamiltonians .
appendix a provides three examples from the ordinary quantum mechanics , the harmonic oscillator , the radial oscillator , the darboux - pschl - teller potential .
appendix b is for the four examples from the discrete quantum mechanics , the hamiltonians of the meixner - pollaczek , the continuous hahn , the wilson and the askey - wilson polynomials @xcite , which are known to reduce to the hermite , the laguerre and the jacobi polynomials in certain limits , respectively .
let us start with a generic one - dimensional quantum mechanical ( qm ) system having discrete semi - infinite energy levels only : @xmath4 here we have chosen the constant part of the hamiltonian so that the groundstate energy is zero .
then the hamiltonian is _ positive semi - definite _ and can be factorised , @xmath5 here a real and smooth function @xmath6 is called a _ prepotential _ and it parametrises the groundstate wavefunction @xmath7 , which has _ no node _ and can be chosen real and positive : @xmath8 it is trivial to verify @xmath9 in one dimension all the energy levels are non - degenerate . by construction
all the eigenfunctions are square - integrable and orthogonal with each other and form a complete basis of the hilbert space : @xmath10 where @xmath11 is the set of non - negative integers @xmath12 .
it is well - known that the @xmath13-th excited wavefunction @xmath14 has @xmath13 nodes in the interior . for simplicity
we choose all the eigenfunctions to be real .
a few exactly solvable examples are given in appendix .
let us choose a set of @xmath3 distinct non - negative integers @xmath15 , satisfying the condition @xmath16 this condition means that the set @xmath17 consists of several clusters , each containing an _ even number _ of _ contiguous _ integers @xmath18 where @xmath19 .
if @xmath20 for the lowest lying cluster , it could contain an even or odd number of contiguous integers .
the set @xmath17 specifies the energy levels to be _
deleted_. this simply reflects the fact that no singularity arises when two neighbouring levels are deleted . in the ordinary qm , the zeros of the two neighbouring eigenfunctions @xmath21 and @xmath22 interlace with each other .
this fact is essential for the non - singularity of the potential after deletion .
see adler s paper @xcite for a proof .
the situation is essentially the same in the discrete qm .
however , in the dqm to be discussed in the subsequent section , due to the lack of general theorem , the interlacing of the zeros of the two neighbouring eigenfunctions @xmath21 and @xmath22 must be verified for each specific hamiltonian . deleting an arbitrary number of contiguous energy levels starting from the groundstate ( @xmath20 )
is achieved by the original crum s theorem @xcite .
next we will construct hamiltonian systems corresponding to the successive deletions @xmath23 ( and @xmath24 , @xmath25 , etc . )
step by step , algebraically .
it should be noted that some quantities in the intermediate steps could be singular . for given @xmath26
the first hamiltonian @xmath27 can be expressed in two different ways : @xmath28 unless @xmath29 , @xmath30 is singular due to the zeros of @xmath31 .
it is very important to note that @xmath32 in is a ` formal adjoint ' of @xmath33 .
we stick to this notation , since the algebraic structure of various expressions appearing in the deletion processes , from to , are best described by using the ` formal adjoint ' .
these define a new hamiltonian system @xmath34 with the ` eigenfunctions ' @xmath35 note that the energy level @xmath26 is now deleted , @xmath36 , from the set of ` eigenfunctions ' @xmath37 of the new hamiltonian @xmath38 .
suppose we have determined @xmath39 and @xmath40 with @xmath41 deletions .
they have the following properties @xmath42 we have also @xmath43 next we will define a new hamiltonian system with one more deletion of the level @xmath44
. we can show @xmath45 these determine a new hamiltonian system with @xmath46 deletions : @xmath47 after deleting all the @xmath48 energy levels , the resulting hamiltonian system @xmath49 , @xmath50 , etc has the following form : @xmath51 now that @xmath52 has the lowest energy level @xmath53 : @xmath54 with the groundstate wavefunction @xmath55 @xmath56 as usual the hamiltonian system can be expressed simply in terms of the groundstate wavefunction @xmath55 , which we will denote by new symbols @xmath57 , @xmath58 , etc : @xmath59 as shown by krein - adler @xcite , the results can be expressed succinctly : @xmath60(x ) } { \text{w}[\phi_{d_1},\phi_{d_2},\ldots,\phi_{d_\ell}](x ) } \quad(n\in\mathbb{z}_+\backslash\mathcal{d } ) , \label{qmphib1bsn}\\ & \bar{u}(x)\equiv u_{d_1,\ldots , d_\ell}(x)=u(x ) -2\frac{d^2}{dx^2}\bigl ( \log\text{w}\,[\phi_{d_1},\phi_{d_2},\ldots,\phi_{d_\ell}](x)\bigr ) \quad(\ell\geq 0 ) , \label{ub1bs}\end{aligned}\ ] ] in which the wronskian determinant is defined by @xmath61(x ) { \stackrel{\text{def}}{=}}\det\bigl(\frac{d^{j-1}f_k(x)}{dx^{j-1}}\bigr)_{1\leq j , k\leq n}.\ ] ] for @xmath62 , we set @xmath63(x)=1 $ ] . in deriving the determinant formulas and use
is made of the properties of the wronskian @xmath64(x ) = g(x)^n\text{w}[f_1,f_2,\ldots , f_n](x ) , \label{wformula1}\\ & \text{w}\bigl[\text{w}[f_1,f_2,\ldots , f_n , g ] , \text{w}[f_1,f_2,\ldots , f_n , h]\,\bigr](x){\nonumber\\}&=\text{w}[f_1,f_2,\ldots , f_n](x)\ , \text{w}[f_1,f_2,\ldots , f_n , g , h](x ) \qquad(n\geq 0 ) .
\label{wformula2}\end{aligned}\ ] ] let us note that @xmath65 and @xmath66 are symmetric with respect to @xmath67 , and thus @xmath68 is independent of the order of @xmath69 .
let us state adler s theorem again ; _ if the set of deleted energy levels @xmath67 satisfy the condition , the hamiltonian @xmath70 with is well - defined and hermitian , and its complete set of eigenfunctions are given by . _
crum s theorem corresponds to the choice @xmath71 , and the resulting lowest energy level is @xmath72 and there is no deleted energy levels above the new groundstate . in this subsection
we consider the typical case of shape invariant systems in which the eigenfunctions consist of the orthogonal polynomials @xmath73 : @xmath74 in which @xmath75 is called the _
sinusoidal coordinate_. as shown in detail in the examples in appendix a , @xmath76 for the harmonic oscillator ( the hermite polynomials ) , @xmath77 for the radial oscillator ( the laguerre polynomials ) and @xmath78 for the darboux - pschl - teller potential ( the jacobi polynomials ) .
the groundstate wavefunction @xmath7 provides the orthogonality weight function @xmath79 in this case , the modification of crum s theorem produces the eigenfunctions @xmath80 which again consist of polynomials in @xmath75 . by using and @xmath81(x )
= \bigl(\tfrac{d\eta(x)}{dx}\bigr)^{\frac12n(n-1 ) } \text{w}[f_1,f_2,\ldots , f_n](\eta(x ) ) , \label{wformula3}\ ] ] we obtain a simple expression of the eigenfunctions @xmath82(\eta(x ) ) } { \text{w}[p_{d_1},p_{d_2},\ldots , p_{d_\ell}](\eta(x))}.\ ] ] this simply means that the resulting eigenfunctions are again polynomials in @xmath75 : @xmath83(\eta(x))},\quad \mathcal{p}_n(\eta){\stackrel{\text{def}}{=}}\text{w}[p_{d_1},p_{d_2},\ldots , p_{d_\ell},p_n](\eta ) , \label{polyphi0}\end{aligned}\ ] ] satisfying the orthogonality relation @xmath84 let us emphasise that @xmath13 is not the _ degree _ in @xmath85 and by construction , @xmath3 members are missing : @xmath86 .
therefore these polynomials can not be called _ exceptional orthogonal polynomials _ @xcite .
let us begin with a few general remarks on the one - dimensional discrete qm with pure imaginary shifts .
see @xcite for the general introduction to the discrete quantum mechanics with pure imaginary shifts and @xcite for the crum s theorem in the discrete qm . in the discrete qm , the dynamical variables are , as in the ordinary qm , the coordinate @xmath87 , which takes value in an infinite or a semi - infinite or a finite range of the real axis and the canonical momentum @xmath88 , which is realised as a differential operator @xmath89 .
since the momentum operator appears in exponentiated forms @xmath90 , @xmath91 , in a hamiltonian , it causes finite pure imaginary shifts in the wavefunction @xmath92 .
this requires the wavefunction as well as other functions appearing in the hamiltonian to be _ analytic _ in @xmath87 within a certain complex domain including the physical region of the coordinate .
let us introduce the @xmath93-operation on an analytic function , @xmath94 . if @xmath95 , @xmath96 , then @xmath97 , in which @xmath98 is the complex conjugation of @xmath99 .
obviously @xmath100 and @xmath101 .
if a function satisfies @xmath102 , we call it a ` real ' function , for it takes real values on the real line . the starting point is again a generic one dimensional discrete quantum mechanical hamiltonian with discrete semi - infinite energy levels only .
again we assume that the groundstate energy is chosen to be zero @xmath103 , so that the hamiltonian is positive semi - definite .
the generic factorised hamiltonian reads @xcite @xmath104 since the @xmath93-operation for @xmath105 , @xmath106 and @xmath107 satisfies @xmath108 they map a ` real ' function to a ` real ' function @xmath109 by specifying the function @xmath110 , various explicit examples are obtained @xcite .
a few exactly solvable examples are given in appendix .
the corresponding schrdinger equation @xmath111 is a _ difference equation _
@xmath112 instead of differential in the ordinary qm
. although this equation looks rather complicated , the equation for the polynomial eigenfunctions has a familiar form of difference equations . again
the groundstate wavefunction @xmath7 is determined as a zero mode of @xmath113 , @xmath114 ( @xmath115 @xmath116 ) , namely , @xmath117 this dictates how the ` phase ' of the potential function @xmath118 is related to that of the groundstate wavefunction @xmath119 . here
we also assume that the groundstate wavefunction @xmath7 has no node and chosen to be real and positive for real @xmath87 .
due to the lack of generic theorems in the theory of difference equations , let us assume that all the energy levels are non - degenerate and that all the eigenfunctions are square - integrable and orthogonal with each other and form a complete basis of the hilbert space : @xmath120 in most explicit examples these statements can be verified straightforwardly . for simplicity
we choose all the eigenfunctions to be real on the real axis @xmath121 , which is made possible by .
the formulation of the modified crum s theorem in the discrete quantum mechanics goes almost parallel to that in the ordinary quantum mechanics .
again the presentation is purely algebraic .
let us note that various quantities in intermediate steps might have singularities and hamiltonians might not be hermitian .
we choose a set of distinct non - negative integers @xmath122 , satisfying the condition as before .
first let us note that the hamiltonian @xmath27 can be rewritten by incorporating the level @xmath26 as : @xmath123 these define a new hamiltonian system @xmath124 note that the energy level @xmath26 is now deleted , @xmath36 , from the set of ` eigenfunctions ' @xmath37 of the new hamiltonian @xmath38 .
suppose we have determined @xmath39 and @xmath40 with @xmath41 deletions .
they have the following properties @xmath125 where @xmath126 .
we have also @xmath43 next we will define a new hamiltonian system with one more deletion of the level @xmath44 .
we can show the following : @xmath127 these determine a new hamiltonian system with @xmath46 deletions : @xmath128 where @xmath129 . after deleting all the @xmath48 energy levels ,
the resulting hamiltonian system @xmath49 , @xmath50 , etc has the following form : @xmath130 now that @xmath52 has the lowest energy level @xmath53 : @xmath131 with the groundstate wavefunction @xmath55 @xmath132 then the hamiltonian system can be expressed simply in terms of the groundstate wavefunction @xmath55 , which we will denote by new symbols @xmath133 , @xmath134 , etc : @xmath135 the discrete counterpart of the determinant formulas requires a deformation of the wronskian , the casorati determinant , which has a good limiting property : @xmath136(x ) { \stackrel{\text{def}}{=}}i^{\frac12n(n-1 ) } \det\bigl(f_k(x+i\tfrac{n+1 - 2j}{2}\gamma)\bigr)_{1\leq j , k\leq n},\\ & \lim_{\gamma\to 0}\gamma^{-\frac12n(n-1 ) } \text{w}_{\gamma}[f_1,f_2,\ldots , f_n](x ) = \text{w}\,[f_1,f_2,\ldots , f_n](x),\end{aligned}\ ] ] ( for @xmath62 , we set @xmath137(x)=1 $ ] . ) .
it satisfies @xmath136^*(x ) = \text{w}_{\gamma}[f_1^*,\ldots , f_n^*](x),\\ & \text{w}_{\gamma}[gf_1,gf_2,\ldots , gf_n ] = \prod_{j=1}^ng(x+i\tfrac{n+1 - 2j}{2}\gamma)\cdot \text{w}_{\gamma}[f_1,f_2,\ldots , f_n](x ) , \label{dwformula1}\\ & \text{w}_{\gamma}\bigl[\text{w}_{\gamma}[f_1,f_2,\ldots , f_n , g ] , \text{w}_{\gamma}[f_1,f_2,\ldots , f_n , h]\,\bigr](x){\nonumber\\}&=\text{w}_{\gamma}[f_1,f_2,\ldots , f_n](x)\ , \text{w}_{\gamma}[f_1,f_2,\ldots , f_n , g , h](x ) \quad(n\geq 0 ) .
\label{dwformula2}\end{aligned}\ ] ] by using the casorati determinant we obtain ( @xmath138 ) @xmath139(x ) } { \text{w}_{\gamma}\,[\phi_{d_1},\ldots,\phi_{d_\ell } ] ( x - i\tfrac{\gamma}{2 } ) } , \label{phib1bsn}\\ & \bar{v}(x)\equiv v_{d_1\,\ldots\,d_\ell\,\mu}(x)= \sqrt{v(x - i\tfrac{\ell}{2}\gamma)v^*(x - i\tfrac{\ell+2}{2}\gamma)}{\nonumber\\}&\phantom{v_{d_1\,\ldots\,d_\ell\,\mu}(x)=}\times \frac{\text{w}_{\gamma}\,[\phi_{d_1},\ldots,\phi_{d_\ell } ] ( x+i\tfrac{\gamma}{2 } ) } { \text{w}_{\gamma}\,[\phi_{d_1},\ldots,\phi_{d_\ell } ] ( x - i\tfrac{\gamma}{2})}\ , \frac{\text{w}_{\gamma}\,[\phi_{d_1},\ldots,\phi_{d_{\ell}},\phi_{\mu } ] ( x - i\gamma ) } { \text{w}_{\gamma}\,[\phi_{d_1},\ldots,\phi_{d_{\ell}},\phi_{\mu}](x)}. \label{vb1bs+1}\end{aligned}\ ] ] we also have ( @xmath138 ) @xmath140 ( x - i\tfrac{\gamma}{2 } ) } { \text{w}_{\gamma}\,[\phi_{d_1},\ldots,\phi_{d_\ell } ] ( x+i\tfrac{\gamma}{2})}. \label{prodv}\ ] ] therefore @xmath141 and @xmath66 are symmetric with respect to @xmath67 , and @xmath142 is independent of the order of @xmath69 . let us state the discrete qm analogue of adler s theorem ; _ if the set of deleted energy levels @xmath143 satisfy the condition , the modified hamiltonian is given by @xmath144 with the potential function given by and its eigenfunctions are given by _ .
the discrete qm version of crum s theorem @xcite corresponds to the choice @xmath71 and the new groundstate is at the level @xmath72 and there is no vacant energy level above that . due to the lack of generic theorems in the theory of difference equations ,
the hermiticity of the resulting hamiltonian @xmath133 and the non - singularity of the eigenfunctions @xmath145 can not be proved categorically for the discrete qm , even when the condition is satisfied by the deleted levels .
see appendix a of @xcite for a detailed discussion of the self - adjointness of the hamiltonians in discrete qm . it should be stressed that in most practical cases , in particular , in the cases of polynomial eigenfunctions , the hermiticity of the hamiltonian @xmath133 and non - singularity of the eigenfunctions @xmath80 are satisfied . in this subsection
we consider the typical case of shape invariant systems in which the eigenfunctions consist of the orthogonal polynomials @xmath73 : @xmath146 in which @xmath75 is called the _ sinusoidal coordinate _ , see .
the groundstate wavefunction @xmath7 provides the orthogonality weight function @xmath147 the difference equation for @xmath73 looks much simpler than the schrdinger equation : @xmath148 for the explicit forms of @xmath110 , see for example , these are the equations that determine the hypergeometric orthogonal polynomials , _ e.g. _ the meixner - pollaczek ( mp ) , the continuous hahn ( ch ) , the wilson ( w ) and the askey - wilson ( aw ) polynomials .
in fact , the above form of the _ difference equation _ is independent of the fact that @xmath149 is a polynomial or not .
it is obtained simply by the similarity transformation of the hamiltonian in terms of the groundstate wavefunction @xmath7 : @xmath150 in the case of polynomial eigenfunctions , the modification of crum s theorem produces the eigenfunctions @xmath80 which again consist of polynomials in @xmath75 . by using the property we have @xmath151(x )
= \prod_{j=1}^{\ell } \phi_0\bigl(x+i\tfrac{\ell+1 - 2j}{2}\gamma\bigr)\cdot \text{w}_{\gamma}[\check{p}_1,\ldots,\check{p}_{\ell}](x),\\ & \text{w}_{\gamma}[\phi_1,\ldots,\phi_{\ell},\phi_n](x ) = \prod_{j=1}^{\ell+1 } \phi_0\bigl(x+i\tfrac{\ell+2 - 2j}{2}\gamma\bigr)\cdot \text{w}_{\gamma}[\check{p}_1,\ldots,\check{p}_{\ell},\check{p}_n](x),\end{aligned}\ ] ] where @xmath152 .
corresponding to the formula in the ordinary qm , we have @xmath153(x ) = \varphi_n(x)\times\bigl(\text{polynomial in $ \eta(x)$}\bigr),\end{aligned}\ ] ] in which @xmath154 is defined in ( and @xmath155 is defined in ) .
these simply mean that the resulting eigenfunctions @xmath80 are again polynomials in @xmath75 : @xmath156 in which @xmath157 and @xmath158 are certain polynomials in @xmath75 defined by @xmath159 = \varphi_{\ell+1}(x)\times\mathcal{p}_n(\eta(x)),\quad \text{w}_{\gamma}[\check{p}_{d_1},\ldots,\check{p}_{d_\ell } ] = \varphi_{\ell}(x)\times\mathcal{q}(\eta(x ) ) .
\label{calpqdef}\ ] ] the polynomials @xmath160 form a complete basis of the hilbert space and satisfy the orthogonality relations @xmath84 let us emphasise that @xmath13 is the level of the original eigenfunction and not the _ degree _ in @xmath85 .
the degree of @xmath161 depends on the set @xmath17 , and it can be calculated explicitly from . by construction @xmath3
members are missing : @xmath86 .
therefore these polynomials can not be called _ exceptional orthogonal polynomials _ @xcite .
theory of exactly solvable discrete qm is less developed than that of the ordinary qm . up to date , the known exactly solvable discrete quantum systems are all shape invariant @xcite and in one to one correspondence with the known ( @xmath2)-hypergeometric orthogonal polynomials @xcite@xcite .
small progress was made in this direction @xcite by introducing several new sinusoidal coordinates for the construction of new types of exactly solvable hamiltonians .
roughly speaking , this approach attempts to create the discrete analogues of various morse type potentials and the soliton potentials . in this paper
we pursue another direction ; to construct infinitely many exactly solvable quantum systems by _ deforming _ the known exactly solvable one . in the ordinary qm , the modification of crum s theorem @xcite due to krein - adler @xcite allows to produce an essentially iso - spectral hamiltonian by deleting a finite number of energy levels from the original system .
the set of deleted level must satisfy certain condition , but there are infinitely many possible deletions leading to infinitely many exactly solvable systems starting from a known one .
the discrete analogue of adler s modification is presented in this paper in parallel with the original version , since the algebraic structure is common .
we also comment on the practical cases when the eigenfunctions consist of orthogonal polynomials .
the eigenfunctions of the resulting system also consist of orthogonal polynomials .
but certain members of these polynomials are missing due to the deletion . very special and simple examples , in which all the excited states from the first to the @xmath3-th are deleted ( see fig.2 ) ,
are presented explicitly in appendix . as will be commented
shortly , these examples were instrumental for the discovery of the infinitely many shape invariant systems and the corresponding infinitely many exceptional orthogonal polynomials @xcite . in the ordinary qm
, the corresponding prepotential has a very simple form : @xmath162 to be compared with the prepotential for the hamiltonian of the @xmath3-th exceptional orthogonal polynomial ( 14 ) and ( 28 ) of @xcite : @xmath163 in the discrete qm , the corresponding formula is @xmath164 to be compared with the corresponding formula for the hamiltonian of the @xmath3-th exceptional orthogonal polynomial ( 30 ) of @xcite : @xmath165 the addition ( multiplication ) of the deforming polynomial with the shifted parameters @xmath166 would achieve the shape invariance . since the harmonic oscillator has no shiftable parameter , we have @xmath167 .
this also ` explains ' non - existence of exceptional hermite polynomials .
in contrast to the hermite polynomial , the continuous hahn polynomial has four real parameters .
we can construct the corresponding exceptional continuous hahn polynomials with three real parameters , which will be reported elsewhere .
the actual function forms of the deforming polynomial @xmath168 in appendix are not the same as those for the exceptional orthogonal polynomials . for
the ordinary qm examples , see vs. ( 13 ) and ( 27 ) in @xcite and for the discrete qm examples , see vs. ( 64 ) and ( 78 ) in @xcite .
but they share some interesting features . before closing this section ,
let us remark that the present modification of crum s theorem is applicable to the hamiltonian systems of various species of the infinite family of exceptional orthogonal polynomials @xcite , as well as to those of the classical orthogonal polynomials including the wilson and the askey - wilson polynomials .
we thank reido kobayashi for useful comments and discussion . r.s .
is supported in part by grant - in - aid for scientific research from the ministry of education , culture , sports , science and technology ( mext ) , no.19540179 .
thanks mext for the research studentship .
* appendix * in appendix we present very special and simple examples of an application of adler s theorem , in which the eigenstates @xmath169 are deleted . in other words ,
@xmath170@xmath171 , that is , the modified groundstate level is the same as that of the original theory @xmath172 .
the situation is illustrated in fig.2 , which should be compared with fig.1 , depicting the generic case discussed in sections [ genordqm ] and [ gendqm ] .
figure1 : generic case figure2 : special case the black circles denote the energy levels , whereas the white circles denote _ deleted _ energy levels .
we write @xmath173 , @xmath174 , @xmath175 , @xmath176 etc . as @xmath177 , @xmath178 , @xmath179 , @xmath180 etc .
this hamiltonian @xmath181 is non - singular for even @xmath3 but may be singular for odd @xmath3 . since algebraic formulas such as the wronskians and casoratians are valid for even and odd @xmath3 , we present various formulas without restricting to the even @xmath3 .
the original systems are shape invariant but the @xmath182-deleted systems @xmath177 are not .
the rightmost vertical line in fig.2 corresponds to the hamiltonian system @xmath183 , which is shape invariant and it is obtained from @xmath177 by one more step of crum s method .
this study helped us to find the new shape invariant systems and exceptional orthogonal polynomials @xcite .
here we apply adler s theorem to the harmonic oscillator , the radial oscillator and the darboux - pschl - teller potential , whose eigenfunctions are described by the classical orthogonal polynomials .
that is , the hermite , laguerre and jacobi polynomials , to be abbreviated as h , l and j , respectively .
these original systems are shape invariant , meaning a very special form of parameter dependence , , . here
we display the parameter dependence explicitly by @xmath184 , which represents the set of the parameters . here
we summarise various properties of the original hamiltonian systems to be compared with the specially modified systems to be presented in [ sec : a2 ] .
let us start with the hamiltonians , schrdinger equations and eigenfunctions ( @xmath185 ) : @xmath186 here @xmath75 is the sinusoidal coordinate , @xmath187 is the prepotential and @xmath188 is the @xmath13-th energy eigenvalue : @xmath189 -\frac12x^2+g\log x&:\text{l}\\[1pt ] g\log\sin x+h\log\cos x&:\text{j } \end{array}\right.\!\ ! , \quad \mathcal{e}_n({\boldsymbol}{\lambda}){\stackrel{\text{def}}{=}}\left\ { \begin{array}{ll } 2n&:\text{h}\\ 4n&:\text{l}\\ 4n(n+g+h)&:\text{j } \end{array}\right.\!\!.\end{aligned}\ ] ] the eigenfunction consists of an orthogonal polynomial @xmath190 , a polynomial of degree @xmath13 in @xmath85 , ( @xmath191 for @xmath192 ) : @xmath193 l_n^{(g-\frac12)}(\eta)&:\text{l}\\[1pt ] p_n^{(g-\frac12,h-\frac12)}(\eta)&:\text{j } \end{array}\right.\!\ ! , \quad c_n({\boldsymbol}{\lambda}){\stackrel{\text{def}}{=}}\left\ { \begin{array}{ll } 2^n&:\text{h}\\ \frac{(-1)^n}{n!}&:\text{l}\\[1pt ] \frac{(n+g+h)_n}{2^nn!}&:\text{j } \end{array}\right.\!\!,\end{aligned}\ ] ] in which @xmath194 is the pochhammer symbol .
shape invariance means @xmath195 or equivalently , @xmath196 the action of @xmath197 and @xmath198 on the eigenfunction is : @xmath199 here the coefficients @xmath200 and @xmath201 are the factors of @xmath188 : @xmath202 the forward and backward shift operators , @xmath203 and @xmath204 , are defined by : @xmath205 and their action on the polynomial is : @xmath206 note that @xmath203 and @xmath204 can also be expressed in terms of @xmath85 only @xcite .
the orthogonality reads @xmath207
\frac{1}{2\,n!}\gamma(n+g+\frac12)&:\text{l}\\[4pt ] \frac{\gamma(n+g+\frac12)\gamma(n+h+\frac12 ) } { 2\,n!(2n+g+h)\gamma(n+g+h)}&:\text{j } \end{array}\right.\!\!.\end{aligned}\ ] ] the prepotential of the modified system is obtained from up to an additive constant : @xmath208 it is a polynomial ( @xmath209 ) deformation of the shape invariant one @xmath210 .
note that the normalization of @xmath168 does not affect the hamiltonian .
the explicit forms of the deforming polynomial @xmath211 will be given in .
for even @xmath3 , the polynomial @xmath209 has no zero in the range of @xmath87 and the modified hamiltonian system is hermitian , which reads : @xmath212 this system is not shape invariant . as mentioned in section [ summary ] , the above form of the deformed prepotential is closely related to that of the exceptional laguerre and jacobi polynomials . a degree @xmath3 polynomial in @xmath85 ,
@xmath211 is defined by @xmath213(\eta;{\boldsymbol}{\lambda } ) { \stackrel{\text{def}}{=}}\prod_{k=1}^{\ell}k!\,c_k({\boldsymbol}{\lambda})\cdot \xi_{\ell}(\eta;{\boldsymbol}{\lambda}),\ ] ] and the explicit forms are : @xmath214 the eigenfunctions are @xmath215(\eta;{\boldsymbol}{\lambda } ) { \stackrel{\text{def}}{=}}\prod_{k=1}^{\ell}k!\,c_k({\boldsymbol}{\lambda})\cdot(-1)^{\ell } p_{\ell , n}(\eta;{\boldsymbol}{\lambda})\quad \bigl(\rightarrow p_{\ell,0}(\eta;{\boldsymbol}{\lambda})=1\bigr).\end{aligned}\ ] ] note that @xmath216 is a polynomial of degree @xmath13 in @xmath85 and @xmath217 and @xmath216@xmath218 for @xmath219 .
we set @xmath220 for @xmath192 . for even @xmath3 , the eigenpolynomial @xmath221 ( @xmath222 )
has @xmath223 zeros in the range of @xmath87 .
the operators @xmath224 and @xmath225 connect the modified system @xmath226 to the shape invariant system @xmath227 with the parameters @xmath228 , which is denoted by the rightmost vertical line in fig.2 . the @xmath13-th level ( @xmath222 ) of the modified system
@xmath229 is _ iso - spectral _ with the @xmath230-th level of the new shape invariant system @xmath231 : @xmath232 here @xmath233 and @xmath234 are defined by @xmath235 b_{\ell , n-1}({\boldsymbol}{\lambda}){\stackrel{\text{def}}{=}}b_{n-1}({\boldsymbol}{\lambda})\times a^{-1 } , \end{array}\quad a=\left\ { \begin{array}{ll } ( -2)^{\ell}(n-\ell)_{\ell}&:\text{h}\\ 1&:\text{l}\\ ( -2)^{-\ell}(n+g+h+1)_{\ell}&:\text{j } \end{array}\right.\!\ ! , \label{qmflnbln}\ ] ] and they factorise @xmath188 , @xmath236 .
the forward and backward shift operators @xmath237 and @xmath238 , which act on the polynomial eigenfunctions , are defined by : @xmath239 their action on the polynomials is ( @xmath222 ) : @xmath240 note that @xmath237 and @xmath238 can be expressed in terms of @xmath85 @xcite . for @xmath222 ,
the above relation provides a simple formula of the modified eigenpolynomial @xmath216 in terms of @xmath211 and the original eigenpolynomial @xmath241 : @xmath242 in which the coefficient @xmath243 is given by @xmath244 the orthogonality relation for even @xmath3 is : @xmath245 a few historical remarks are in order .
dubov et al @xcite derived in 1992 an exactly solvable hamiltonian system of a deformed harmonic oscillator , which corresponds to the @xmath246 case of this appendix .
their paper , written about two years before adler s , relied on rather heuristic arguments .
recently quesne @xcite derived exactly solvable and non - shape invariant systems of deformed radial oscillator and deformed dpt potential , both are called type iii .
these results again correspond to the @xmath246 cases of the radial oscillator and the dpt potential of this appendix .
here we apply adler s theorem to the shape invariant , therefore solvable , systems whose eigenfunctions are described by the orthogonal polynomials ; the meixner - pollaczek ( we set the parameter @xmath247 ) , continuous hahn , wilson and askey - wilson polynomials @xcite , to be abbreviated as mp , ch , w and aw , respectively .
see @xcite and @xcite for the discrete qm treatment of these polynomials .
here we summarise various properties of the original hamiltonian systems to be compared with the specially modified systems to be presented in [ sec : b2 ] .
let us start with the hamiltonians , schrdinger equations and eigenfunctions ( @xmath185 ) : @xmath248 the set of parameters @xmath184 are @xmath249 where @xmath250 .
the the sinusoidal coordinate @xmath75 is , @xmath251 the potential function @xmath252 and energy eigenvalue @xmath188 are @xmath253 \bigl(2ix(2ix+1)\bigr)^{-1}\prod_{j=1}^4(a_j+ix)&:\text{w}\\[2pt ] \bigl((1-e^{2ix})(1-qe^{2ix})\bigr)^{-1}\prod_{j=1}^4(1-a_je^{ix } ) & : \text{aw } \end{array}\right.\!\ ! , \label{vform}\\ & \mathcal{e}_n({\boldsymbol}{\lambda}){\stackrel{\text{def}}{=}}\left\ { \begin{array}{lll } 2n&&:\text{mp}\\
n(n+b_1 - 1),&b_1{\stackrel{\text{def}}{=}}a_1+a_2+a_1^*+a_2^*&:\text{ch}\\[1pt ] n(n+b_1 - 1),&b_1{\stackrel{\text{def}}{=}}a_1+a_2+a_3+a_4&:\text{w}\\[1pt ] ( q^{-n}-1)(1-b_4q^{n-1}),&b_4{\stackrel{\text{def}}{=}}a_1a_2a_3a_4&:\text{aw } \end{array}\right.\!\!.\end{aligned}\ ] ] the eigenfunction is described by the orthogonal polynomial @xmath190 , a polynomial of degree @xmath13 in @xmath85 : @xmath254 w_n(\eta;a_1,a_2,a_3,a_4)\!\!&:\text{w}\\[1pt ] p_n(\eta;a_1,a_2,a_3,a_4|q)\!\!&:\text{aw } \end{array}\right.\!\ ! , \
\ c_n({\boldsymbol}{\lambda}){\stackrel{\text{def}}{=}}\left\ { \begin{array}{ll } \frac{1}{n!}\,2^n\!\!&:\text{mp}\\[1pt ] \frac{1}{n!}\,(n+b_1 - 1)_n\!\!&:\text{ch}\\[1pt ] ( -1)^n(n+b_1 - 1)_n\!\!&:\text{w}\\[1pt ] 2^n(b_4q^{n-1}\,;q)_n\!\!&:\text{aw } \end{array}\right.\!\!,\end{aligned}\ ] ] in which @xmath255 is the @xmath2-pochhammer symbol .
we set @xmath191 for @xmath192 .
the shape invariance relations involve one more parameter @xmath256 : @xmath257 ( \frac12,\frac12)&:\text{ch}\\[1pt ] ( \frac12,\frac12,\frac12,\frac12)&:\text{w,\,aw } \end{array}\right.\!\ ! , \quad \kappa{\stackrel{\text{def}}{=}}\left\ { \begin{array}{ll } 1&:\text{mp,\,ch,\,w}\\ q^{-1}&:\text{aw } \end{array}\right.\!\!,\end{aligned}\ ] ] or equivalently , @xmath258 the groundstate wavefunction @xmath259 is determined by @xmath260 and its explicit forms are : @xmath261 \sqrt{\gamma(a_1+ix)\gamma(a_2+ix)\gamma(a_1^*-ix)\gamma(a_2^*-ix ) } & : \text{ch}\\[2pt ] \sqrt{(\gamma(2ix)\gamma(-2ix))^{-1 } \prod_{j=1}^4\gamma(a_j+ix)\gamma(a_j - ix)}&:\text{w}\\[2pt ] \sqrt{(e^{2ix}\,;q)_{\infty}(e^{-2ix}\,;q)_{\infty } \prod_{j=1}^4(a_je^{ix}\,;q)_{\infty}^{-1}(a_je^{-ix}\,;q)_{\infty}^{-1 } } & : \text{aw } \end{array}\right.\!\!.\ ] ] we introduce an auxiliary function @xmath155 with the properties : @xmath262 the sinusoidal coordinate @xmath75 has the following properties : @xmath263 2\eta(x)\cosh\tfrac{k\gamma}{2}&:\text{aw } \end{array}\right.\!\!,\\ & \eta(x - i\tfrac{k\gamma}{2})\eta(x+i\tfrac{k\gamma}{2 } ) = \left\ { \begin{array}{ll } \eta(x)^2+\tfrac14k^2&:\text{mp,\,ch}\\ \bigl(\eta(x)+\tfrac14k^2\bigr)^2&:\text{w}\\[1pt ] \eta(x)^2+\sinh^2\tfrac{k\gamma}{2}&:\text{aw } \end{array}\right.\!\!. \label{etaprod}\end{aligned}\ ] ] these mean that for a polynomial @xmath264 in @xmath85 , @xmath265 is another polynomial in @xmath75 .
the action of @xmath197 and @xmath198 on the eigenfunctions is @xmath199 the factors of the energy eigenvalue , @xmath200 and @xmath201 , @xmath266 , are given by @xmath267 the forward and backward shift operators @xmath203 and @xmath204 are defined by @xmath268 and their action on the polynomials is @xmath206 the orthogonality relation is @xmath269 2\pi n!\,(n+b_1 - 1)_n\prod_{1\leq i < j\leq 4}\gamma(n+a_i+a_j)\cdot \gamma(2n+b_1)^{-1}&:\text{w}\\[2pt ] 2\pi(b_4q^{n-1};q)_n(b_4q^{2n};q)_{\infty}(q^{n+1};q)_{\infty}^{-1 } \prod_{1\leq i < j\leq 4}(a_ia_jq^n;q)_{\infty}^{-1}&:\text{aw } \end{array}\right.\!\!.\end{aligned}\ ] ] the potential function , the hamiltonian and the schrdinger equation of the modified system are : @xmath270 the explicit forms of the deforming polynomial @xmath211 will be given in .
for even @xmath3 , @xmath209 has no zero in the rectangular domain in the complex @xmath87 plane , @xmath271 , @xmath272 .
note that the normalization of @xmath168 does not affect @xmath177 .
this system is not shape invariant .
the second line of the expression for @xmath273 , , is obtained from the first line by repeated use of the formula of the auxiliary function @xmath274 . as mentioned in section [ summary ] , this form of the deformed potential function is closely related to that of the exceptional wilson and askey - wilson polynomials .
it is convenient to introduce an auxiliary function @xmath275 : @xmath276}\prod_{k=1}^{\ell-2 } \bigl(\varphi(x - i\tfrac{k}{2}\gamma)\varphi(x+i\tfrac{k}{2}\gamma ) \bigr)^{[\frac{\ell - k}{2 } ] } , \label{varphildef}\ ] ] where @xmath277 $ ] denotes the greatest integer not exceeding @xmath87 .
note that @xmath278 + 2\sum_{k=1}^{\ell-2}[\frac{\ell - k}{2 } ] = \frac12\ell(\ell-1)$ ] .
the deforming polynomial @xmath211 is defined by @xmath279(x;{\boldsymbol}{\lambda } ) \qquad\quad\bigl ( \check{p}_n(x;{\boldsymbol}{\lambda}){\stackrel{\text{def}}{=}}p_n(\eta(x);{\boldsymbol}{\lambda})\bigr){\nonumber\\}&{\stackrel{\text{def}}{=}}\prod_{k=1}^{\ell}c_k({\boldsymbol}{\lambda})\cdot \varphi_{\ell}(x ) \xi_{\ell}(\eta(x);{\boldsymbol}{\lambda } ) \times\left\ { \begin{array}{ll } \prod_{k=1}^{\ell}k!\!\!&:\text{mp,\,ch,\,w}\\ \prod_{k=1}^{\ell}\prod_{j=1}^k\sinh\frac{-j\gamma}{2}\!\!&:\text{aw } \end{array}\right.\!\!.\end{aligned}\ ] ] as in the ordinary qm cases , it is expressed in terms of the polynomial @xmath280 of the original system with shifted parameters : @xmath281 note that @xmath282 and @xmath283 for the mp , w and aw cases .
the eigenfunctions are @xmath284(x;{\boldsymbol}{\lambda } ) { \stackrel{\text{def}}{=}}\prod_{k=1}^{\ell}c_k({\boldsymbol}{\lambda})\cdot \varphi_{\ell+1}(x)(-1)^{\ell } p_{\ell , n}(\eta(x);{\boldsymbol}{\lambda})\\ & \qquad\qquad\qquad\qquad\qquad\qquad \times\left\ { \begin{array}{ll } \prod_{k=1}^{\ell}k!&\!\!:\text{mp,\,ch,\,w}\\ \prod_{k=1}^{\ell}\prod_{j=1}^k\sinh\frac{-j\gamma}{2}&\!\!:\text{aw } \end{array}\right . \bigl(\rightarrow p_{\ell,0}(\eta;{\boldsymbol}{\lambda})=1\bigr ) .
\nonumber\end{aligned}\ ] ] for even @xmath3 , @xmath221 ( @xmath222 ) has @xmath223 zeros in the range of @xmath87 .
note that @xmath216 is a polynomial of degree @xmath13 in @xmath85 and @xmath217 and @xmath220 for @xmath219 .
we set @xmath220 for @xmath192 .
the operators @xmath224 and @xmath225 connect the modified system @xmath285 to the shape invariant system @xmath286 with the parameters @xmath228 , which is denoted by the rightmost vertical line in fig.2 . the @xmath13-th level ( @xmath222 ) of the modified system
@xmath229 is _ iso - spectral _ with the @xmath230-th level of the new shape invariant system @xmath231 : @xmath232 here , @xmath233 and @xmath234 are the factors of the energy eigenvalue , @xmath236 , and are defined by @xmath235 b_{\ell , n-1}({\boldsymbol}{\lambda}){\stackrel{\text{def}}{=}}b_{n-1}({\boldsymbol}{\lambda})\times a^{-1 } , \end{array}\quad a=\left\ { \begin{array}{ll } 2^{\ell}&:\text{mp}\\ ( b_1+n)_{\ell}&:\text{ch}\\ ( -1)^{\ell}(n-\ell)_{\ell}(b_1+n)_{\ell}&:\text{w}\\ q^{-\frac12\ell n}(q^{n-\ell};q)_{\ell}(b_4q^n;q)_{\ell}&:\text{aw } \end{array}\right.\!\!. \label{dqmflnbln}\ ] ] the forward and backward shift operators @xmath237 and @xmath238 which act on the polynomial eigenfunctions , are defined by : @xmath287 and their action on the polynomials is @xmath288 for @xmath222 ,
the above formula provides a simple formula of the modified eigenpolynomial @xmath216 in terms of @xmath211 and the original eigenpolynomial @xmath241 : @xmath289 the orthogonality relation for even @xmath3 is : @xmath290 s.odake and r.sasaki , `` unified theory of annihilation - creation operators for solvable ( ` discrete ' ) quantum mechanics , '' j. math . phys .
* 47 * ( 2006 ) 102102 ( 33pages ) , exact solution in the heisenberg picture and annihilation - creation operators , " phys . lett .
* b641 * ( 2006 ) 112 - 117 .
h.risken , _ the fokker - planck equation _
( 2nd . ed . ) , springer - verlag , berlin , ( 1996 ) .
d.gmez-ullate , n.kamran and r.milson , `` an extension of bochner s problem : exceptional invariant subspaces , '' j. approx theory * 162 * ( 2010 ) 987 - 1006 ; `` an extended class of orthogonal polynomials defined by a sturm - liouville problem , '' j. math .
* 359 * ( 2009 ) 352 - 367 ; exceptional orthogonal polynomials and the darboux transformation , " arxiv:1002.2666[math - ph ] .
c.quesne , `` exceptional orthogonal polynomials , exactly solvable potentials and supersymmetry , '' j. phys . *
a41 * ( 2008 ) 392001 ; b.bagchi , c.quesne and r.roychoudhury , isospectrality of conventional and new extended potentials , second - order supersymmetry and role of pt symmetry , " pramana j. phys .
* 73 * ( 2009 ) 337 - 347 .
s.odake and r.sasaki , `` infinitely many shape invariant potentials and new orthogonal polynomials , '' phys . lett .
* b679 * ( 2009 ) 414 - 417 .
s.odake and r.sasaki , infinitely many shape invariant discrete quantum mechanical systems and new exceptional orthogonal polynomials related to the wilson and askey - wilson polynomials , " phys . lett .
* b682 * ( 2009 ) 130 - 136 .
s.odake and r.sasaki , infinitely many shape invariant potentials and cubic identities of the laguerre and jacobi polynomials , " j. math
* 51 * ( 2010 ) april issue , in press , arxiv:0911.1585[math - ph ] .
s.odake and r.sasaki , `` another set of infinitely many exceptional ( @xmath291 ) laguerre polynomials , '' phys .
* b684 * ( 2010 ) 173 - 176 .
l.ho , s.odake and r.sasaki , `` properties of the exceptional ( @xmath291 ) laguerre and jacobi polynomials , '' dpsu-09 - 7 , yitp-09 - 70 , arxiv:0912.5447[math - ph ] .
s.yu.dubov , v.m.eleonski and n.e.kulagin , `` equidistant spectra of anharmonic oscillators , '' soviet phys .
jetp * 75 * ( 1992 ) 446 - 451 ; chaos * 4 * ( 1994 ) 47 - 53 .
m.g.krein , doklady acad .
cccp , * 113 * ( 1957 ) 970 - 973 ; v..adler , `` a modification of crum s method , '' theor . math . phys .
* 101 * ( 1994 ) 1381 - 1386 .
p.b.abraham and h.e.moses , changes in potentials due to changes in the point spectrum : anharmonic oscillators with exact solutions , " phys . rev .
* a22 * ( 1980 ) 1333 - 1340 ; b.mielnik , factorization method and new potentials with the oscillator spectrum , " j. math . phys .
* 25 * ( 1984 ) 3387 - 3389 ; m.m.nieto , relationship between supersymmetry and the inverse method in quantum mechanics , " phys . lett .
* b145 * ( 1984 ) 208 - 210 .
g.junker and p.roy , conditionally exactly solvable problems and nonlinear algebras , " phys .
* a232 * ( 1997 ) 155 - 161 ; conditionally exactly solvable potentials : a supersymmetric construction method , " annals of phys .
* 270 * ( 1998 ) 155 - 177 .
s.odake and r.sasaki , `` shape invariant potentials in ` discrete ' quantum mechanics , '' j. nonlinear math
* 12 * suppl . 1 ( 2005 ) 507 - 521 ; `` equilibrium positions , shape invariance and askey - wilson polynomials , '' j. math . phys . * 46 * ( 2005 ) 063513 ( 10 pages ) .
s.odake and r.sasaki , `` calogero - sutherland - moser systems , ruijsenaars - schneider - van diejen systems and orthogonal polynomials , '' prog .
theor . phys .
* 114 * ( 2005 ) 1245 - 1260 .
s.odake and r.sasaki , orthogonal polynomials from hermitian matrices , " j. math . phys .
* 49 * ( 2008 ) 053503 ( 43 pages ) .
s.odake and r.sasaki , exactly solvable ` discrete ' quantum mechanics ; shape invariance , heisenberg solutions , annihilation - creation operators and coherent states , " prog .
* 119 * ( 2008 ) 663 - 700 .
a.f.nikiforov , s.k.suslov , and v.b.uvarov , _ classical orthogonal polynomials of a discrete variable _ , springer , berlin , ( 1991 ) .
s.odake and r.sasaki ,
@xmath2-oscillator from the @xmath2-hermite polynomial , " phys . lett . *
b663 * ( 2008 ) 141 - 145 .
p.gaillard and v.b.matveev , wronskian and casorati determinant representation for darboux - pschel - teller potentials and their difference extensions , " j. phys .
* a42 * ( 2009 ) 404009 .
|
crum s theorem in one - dimensional quantum mechanics asserts the existence of an associated hamiltonian system for any given hamiltonian with the complete set of eigenvalues and eigenfunctions .
the associated system is iso - spectral to the original one except for the lowest energy state , which is deleted .
a modification due to krein - adler provides algebraic construction of a new complete hamiltonian system by deleting a finite number of energy levels . here
we present a discrete version of the modification based on the crum s theorem for the ` discrete ' quantum mechanics developed by two of the present authors .
yukawa institute kyoto + dpsu-10 - 1 + yitp-10 - 15 + * modification of crum s theorem for ` discrete ' quantum mechanics + + * * leonor garca - gutirrez@xmath0 , satoru odake@xmath1 and ryu sasaki@xmath0 * @xmath0 yukawa institute for theoretical physics , + kyoto university , kyoto 606 - 8502 , japan + @xmath1 department of physics , shinshu university , + matsumoto 390 - 8621 , japan
|
neutrino physics is living in a `` golden era '' .
experiments with atmospheric and solar neutrinos have given strong evidence for the existence of flavor oscillations , or ( less likely ) some other form of ` new physics ' beyond the standard model @xcite .
the study of the @xmath2 flavor transitions offer the possibility to obtain information about the neutrino masses and mixing , and the knowledge about these quantities hopefully represents a fundamental window on the physics of the unification .
theorists are busy trying to make sense of the unexpected results that have been obtained so far , on the other hand new experimental studies are planned to measure with greater precision the properties of the @xmath2 flavor transitions .
these future studies , are designed to determine if the @xmath2 flavor evolution is completely described by the 3-flavor mixing model , or if a more complex dynamics ( transitions to additional sterile states , neutrino decay , flavor changing neutral currents ) is necessary , and should measure the entire set of parameters ( 2 squared mass differences , 3 mixing angles and one cp violating phase ) that determine the @xmath2 flavor evolution in space time in the 3flavor scenario . the discovery of @xmath2 oscillations has been performed with natural ( solar and atmospheric ) neutrino sources , however most of the future studies will be performed with artificial neutrinos . in particular , long baseline ( lbl ) accelerator neutrino beams will play a fundamental role .
the first of these projects , the kek to kamioka ( k2k ) @xmath2beam , with @xmath3 gev and @xmath4 km , has already collected data , that support the existence of the @xmath5 transitions as indicated by the atmospheric neutrino data @xcite .
other projects are already funded and in construction ( fermilab to minos @xcite , and cern to gran sasso @xcite , both with @xmath6 km ) , or in the design phase ( the jhf project @xcite ) .
several ` next generation ' projects are also in discussion , involving `` super - beams '' , ( very intense @xmath2beams produced with the traditional technique : @xmath7 ) , or neutrino factories ( muon storage rings that produce extraordinarily well controlled neutrino beams from muon decay : @xmath8 . the @xmath2 energy ( and pathlength ) in the lbl projects is determined by their scientific goals and the existing technical and experimental constraints .
the range of energy @xmath910 ) gev will have crucial importance .
in fact the difference in quantum mechanical phase developed by two mass eigenstates after propagation for a distance @xmath10 is : @xmath11 where @xmath12 is the @xmath2 squared mass difference .
the probability for flavor transitions becomes important when at least one quantum phase difference is of order unity , and this determines the combinations of @xmath10 and @xmath13 that are needed to study the flavor transitions once the order of magnitude of the @xmath14 are known .
the largest squared mass difference , measured in atmospheric @xmath2 experiment : @xmath15 is of order @xmath16 ev@xmath17 .
a quantum mechanical phase difference @xmath18 corresponds to the energy @xmath19 or numerically : @xmath20 \left [ { l \over 250~\;{\rm km } } \right]\ ] ] at this energy one expects a maximum of the @xmath21 oscillation probability for a distance @xmath10 between @xmath2 source and detector .
note that also the evidence for oscillation for atmospheric neutrinos has been obtained with neutrinos of energy @xmath0 gev .
the top panel of fig . 1 shows the energy spectra of different long baseline projects , and for atmospheric neutrinos neutrinos , with the process @xmath22 .
since the threshold for the @xmath23 production is @xmath24 gev , the beam is designed to have a larger average @xmath2 energy . ] .
the flavor composition , and energy spectrum of a @xmath2 beam in a detector ( located at a distance @xmath10 from the source ) is determined from the observation of neutrino interactions , and clearly uncertainties in the knowledge of the neutrino cross sections are a limiting factors in the sensitivity .
there are uncertainties about the absolute value of the cross sections , the ratio @xmath25 between the neutral and charged current interaction rates , the average multiplicity in the final state , the energy and angular distribution of final state lepton ( and hadronic particles ) . depending of the detector type and the scientific question asked
, these uncertainties can have a different importance .
the nuint01 workshop has been dedicated to a discussion of the problem of the neutrino interaction properties , attempting to estimate : ( i ) the size of the existing uncertainties , ( ii ) their probable impact on studies of the fundamental @xmath2 properties , and ( iii ) possible methods to reduce these uncertainties .
a list of relevant questions is : * how well have we measured the @xmath2 cross sections and interaction properties ? *
how good is our theoretical understanding ( and computational capabilities ) of the @xmath2 cross sections and interaction properties ? * how well do we _ need _ to know @xmath26 to determine with the desired accuracy the @xmath2 flavor evolution and mass matrix ? *
what experimental programs can / should be planned in order to improve our knowledge and understanding of the @xmath2 interaction properties ?
the neutrino cross section can be obtained from data , from the formula : @xmath27 where @xmath28 is the number of detected interactions , @xmath29 is the neutrino flux , and @xmath30 is detector efficiency .
this measurement has been performed in the energy region of interest ( @xmath0 gev ) by several experiments@xcite-@xcite .
the most accurate one were performed with bubble chambers filled with hydrogen or deuterium .
a review of this data is given in @xcite .
the uncertainty of @xmath26 is dominated by systematic errors in the determination of the @xmath2 flux .
for example for the data of @xcite ( the 12foot hydrogen / deuterium bubble chamber at the argonne national laboratory ) , the authors estimated an overall 15% uncertainty for the neutrino beam that was produced by a 12.4 gev proton beam interacting on a beryllium target .
the systematic uncertainty was estimated as as a convolution of different factors : uncertainties on the fundamental @xmath31 production cross section ( 5% ) , uncertainties of pion reabsorption in the target ( 5% ) , proton beam intensity ( 2% ) , and smaller contributions .
recent critical analysis @xcite of the uncertainties in the @xmath32nucleus cross sections , related to the calculations of the atmospheric neutrino fluxes have concluded that the uncertainties on the hadronic cross sections are larger ( of order 1520% ) . to be conservative
, it could be safer to consider a systematic larger that what was quoted in the original experiments .
the facilities developed for the long baseline experiments can provide @xmath2 beams significantly more intense that the beams used in the 70 s . in order to use these beams to obtain absolute values for the cross sections it is essential to have good control of the normalization and energy shape of the neutrino beam .
the k2k experiment has already collected in the near detector a very important sample of events ( see section 4 ) , and a problem of considerable interest is to determine how well the @xmath2 beam is known .
the same question clearly applies for the near detector in the fermilab to minos project . in principle
it is possible to determine the quasi
elastic cross section on ( quasi)free nucleons without a precise knowledge of the @xmath2 flux , making use of some theoretical input . if the incident @xmath2 energy is known , from a measurement of the energy and direction of both final state particles in the scattering @xmath33 one can determine both the neutrino energy @xmath13 and the @xmath1 of the reaction reaction has to be performed on a nuclear target ( preferably deuterium ) and nuclear corrections have to be considered . ]
it is therefore possible to determine the _ shape _ of the differential cross section @xmath34 , that to a good approximation can be considered as a function of a single parameter the axial mass @xmath35 ( see next section ) , while the absolute normalization , that is the value of the cross section at @xmath36 can be inferred from @xmath37 decay measurements .
this program has been performed in @xcite ( data from the 7foot hydrogen / deuterium bubble chamber at the brookaven national laboratory ) with the result @xmath38 gev , and in @xcite ( anl 12ft bubble chamber ) with the result @xmath39 gev .
the total cross section ( and therefore the event rate ) is to a good approximation proportional to @xmath40 . in @xcite
the analysis of the event rate results in a lower estimate @xmath41 .
the poor agreement between the two methods suggests the existence of systematic effects .
several other experiments have measured @xmath2 interactions in the energy range of interest using nuclear targets . in the published analysis of these works
very often the results are corrected for nuclear effects , and the data are the expressed as cross sections on free nucleons . since the nuclear effects themselves are difficult to compute ( as discussed in the following ) and the subject of some controversy , some careful considerations are needed to use these results . for practical purposes
it clearly would be very valuable to have available also the results for interactions on nuclear targets , before the nuclear corrections .
an example of this problem can be seen analysing the results of the two @xmath2 experiments @xcite operating with the @xmath2beam at the institute for high energy physics ( ihep ) at serpukhov ( russia ) , that were presented at this workshop .
the skat bubble chamber @xcite has collected data on a set of low multiplicity reactions , namely the final states @xmath42 , @xmath43 , @xmath44 , @xmath45 with an additional non observed ` nuclear spectators ' state .
each one of the considered final states can be obtained as the result of nuclear cascading processes , such as @xmath31 absorption ( @xmath46 ) , @xmath31 production ( @xmath47 , @xmath48 , @xmath49 ; and @xmath31 charge exchange ( for example @xmath50 ) .
the experimentalists have calculated the probabilities for all these processes with montecarlo methods to estimate the rate for exclusive reactions on free nucleons , however these corrections are probably the largest uncertainty in the problem .
only the interactions of neutrinos with nucleons and nuclei are relevant for oscillation studies , since the ratio @xmath51 is of order @xmath52 ; the cross section can naturally be separated into a charged current ( cc ) and a neutral current ( nc ) component : @xmath53 . for low energy neutrinos
it is also natural to decompose the cross section according to the multiplicity : @xmath54 the correct calculation of the quasi elastic and one pion components has great importance in the analysis of existing data since most of the data on atmospheric neutrinos has been obtained with water detectors ( super kamiokande , and before imb and kamiokande ) that , because of relatively poor pattern recognition capabilities , have limited most of their analysis to the so called single
ring events , where only one particle is visible .
this selects quasi elastic scattering with in addition events with one ( or more ) charged pion(s ) below the threshold .
the bottom panel of fig . 1 shows a plot of @xmath55 separated according to the pion multiplicity and compared with some existing data . the importance of the low multiplicity channels is clearly displayed the quasi
elastic cross section can be written in terms of nucleon form factors .
the most general matrix element for a nucleon transition can be written @xcite in terms of six form factors : @xmath56u_n(p ) \\
\end{array}\ ] ] the hypothesis of cvc allows to relate the vector form factors @xmath57 and @xmath58 to the electromagnetic form factors of the proton and neutron ; the contribution of second class currents ( that violate @xmath59parity ) can be assumed to vanish , and one remains with two unknown functions the axial form factor @xmath60 and the pseudoscalar form factor @xmath61 .
the contribution of the pseudoscalar form factor to the cross section is proportional to @xmath62 , therefore this term is important only for @xmath63 , while the axial form factor @xmath60 remains undetermined .
the value of @xmath60 at @xmath64 can be related to the axial coupling measured in @xmath37-decay experiments ( @xmath65 ) , but the @xmath1 dependence of @xmath60 has to be determined experimentally .
an often used parametrization is a dipolar form : @xmath66^{-2 } \label{eq : fa}\ ] ] that depends on the single parameter @xmath35 .
therefore in first approximation the problem of the determination of the quasi elastic cross section can be indentified with the measurement for @xmath67 .
a review of the status of our knowledge of the nucleon form factors was given in @xcite and @xcite .
the kinematics of a final state lepton is described by two variables , that correspond to its energy @xmath68 and angle with respect to the initial neutrino @xmath69 .
the kinematical variables can be chosen as relativistic invariants such as the adimensional scaling variables @xmath70 and @xmath71 ; of particular dynamical importance are the transfer momentum in the interaction : quantities : @xmath72 and the square of the hadronic mass in the final state : @xmath73 the kinematically allowed range of @xmath1 is the interval between @xmath74 and @xmath75 that grows linearly with energy . for @xmath76 ( that is for @xmath77 tev ) the effects of the @xmath78 propagator are small and the cross section @xmath26 grows approximately linearly with energy .
similarly the allowed range of the squared hadronic mass @xmath79 is between the values @xmath80 and @xmath81 . in general @xcite
, the inclusive differential neutrino cross section can be written in terms of 5 structure functions @xmath82 , according to the general formula : @xmath83 \bigg\ } , \end{array}\ ] ] where the two signs for the @xmath84 component correspond to @xmath2 and @xmath85 .
the scaling variables @xmath70 and @xmath71 are defined as @xmath86 and @xmath87 . note that the structure functions @xmath88 and @xmath89 contribute terms proportional to @xmath90 and are therefore of practical interest only in the case of @xmath63 cc interactions ( where they constitute a source of uncertainty ) .
it is well known that in the deep inelastic scattering ( dis ) limit , that corresponds to @xmath1 and @xmath79 large , the structure functions can be related to the parton distribution functions that can be well measured in different experiments , and have a calculable @xmath1 evolution .
for example in leading order of an @xmath91 expansion : @xmath92 therefore in the dis region the neutrino cross section is well understood and under good theoretical control .
however , essentially the entire phase space relevant for atmospheric neutrino contained events , is well outside of the dis region , and this remains true for the k2k , minos to fermilab and jhf projects .
an illustration of this problem is shown in fig . 2 , where we plot the cross section for the scattering @xmath93 at an energy of 3 gev .
at this energy ( that is actually rather high for atmospheric @xmath2 and most lbl experiments ) , the range of possible @xmath78 is small and extends in the region where hadronic resonances are present . the naive use of the dis formula with the pdf s of gluck reya and vogt @xcite results in a smooth curve , that predicts a non vanishing cross section also in the interval @xmath94 .
obviously a realistic calculation must predict a more complex structure , as it is the case for the model of rein and sehgal @xcite whose results are also shown in the figure .
figure 3 is another illustration of the importance of the resonance region in the k2k and fermilab to minos projects .
the delta function for @xmath95 has an area of @xmath96 for k2k and of order 11% for minos ( le beam ) .
a central problem for the calculation of the @xmath2 cross section is the correct description of the low @xmath1 , low @xmath79 kinematical range , where perturbative qcd is not valid .
luckily , in recent times , a large body of highly accurate data on electron
proton scattering in this kinematical region has become available , mostly from the jefferson laboratory ( see @xcite ) .
an example of this data is shown in fig .
[ fig : f2_a ] where is plotted the electromagnetic structure function @xmath97 that in the dis limit in leading order can be expressed in terms of pdf s as : @xmath98\ ] ] one can see that the production of resonances plays a very important role , and that three mass peaks are clearly visible at low @xmath1 .
[ fig : f2_b ] also illustrates the importance of a proper treatment of the relevant effects in the low @xmath1 , region . the physical insight that can be obtained from the electron scattering data
can be applied also to neutrino interactions , even if significant theoretical uncertainties remain .
a concept of great importance in the resonance region is the notion of `` parton hadron '' duality . already in 1971 ,
bloom and gilman @xcite observed empirically , that the scaling structure function @xmath99 that characterizes electromagnetic scattering in the dis ( high @xmath1 region ) , represents a good average over the resonance bumbs of the inclusive data observed at much lower @xmath1 .
the agreement can be significantly improved rescaling the @xmath70 variable for target mass correction and using for example the nachtmann variable @xmath100 .
neutrino detectors are usually composed of heavy nuclei ( the most important nuclear species are oxygen ( sk ) , argon ( icarus ) , iron ( minos ) and lead ( opera ) ) , and in the energy range we are discussing nuclear effects are important and have to be taken properly into account .
schematically one can separate the nuclear effects into two parts : ( i ) the @xmath2 interaction with a bound nucleon , and ( ii ) the propagation of the final state hadrons in the nuclear matter , even if a clear distinction of these two effects is not rigorously valid .
the simplest and most commonly used approximation to describe the interaction with a bound nucleon is the fermi gas model . in this model
the nucleus is described as an ensemble of nucleons with a momentum distribution @xmath101 .
the energy that corresponbds to the 3momentum @xmath102 is @xmath103 where the quantity @xmath104 is the binding energy that can be different for protons and neutrons is also a function of @xmath102 . ] . in the simplest version of the model @xcite
, the initial nucleon momentum distribution has a `` zero temperature '' shape : @xmath105 that vanishes when @xmath106 is larger than the fermi momentum @xmath107 , in other models it has a more complex shape that includes higher momentum tails .
elastic scattering can be simply described as an elementary ( @xmath108 ) process such as @xmath109 where @xmath110 and @xmath111 are off shell initial final states nucleons with 3momenta @xmath102 and @xmath112 ( @xmath113 is the transfer 4momentum ) and binding energies @xmath114 and @xmath115 .
this procedure allows naturally to implement pauli blocking , forbidding the scattering into occupied nucleon states .
this can be done including a factor @xmath116 $ ] in the calculation of the matrix element for the scattering .
some natural questions , that have been discussed at the workshop are : ( i ) how good is the fermi gas model as a description of the nuclear effects ; ( ii ) how large are the expected corrections to the fermi gas model ; ( iii ) what is the best ( or most convenient ) theoretical framework to describe nuclear effects beyond the fermi gas model .
it is clear that conceptually the fermi gas model is a very simple and non
realistic description of the scattering @xcite , it is however not entirely clear how large is the error introduced by this simplified treatment . before , during and after the workshop , it was attempted to produce and compare different `` beyond the fermi gas model '' calculations , ( see the critical discussion by r. seki @xcite in these proceedings ) .
hopefully a generally accepted model will soon emerge from this work .
an example of the results for one of these more sophisticated models ( by g. c and collaborators @xcite ) .
is shown in fig . [
fig : fermi ] .
this model predicts a suppression of the cross section by approximately 15% with respect to the fermi gas model , and a softer spectrum for the final state lepton .
the uncertainties due to nuclear effects become larger when discussing the resonance production processes and the propagation of hadronic particles in the nuclear matter .
it is important to notice that because of pion reabsorption processes the calculation of pion less final states must also include contributions from resonance production .
the k2k experiment has collected data with the neutrino beam obtained from a 12 gev protons beam at the kek accelerator .
the goal of the experiment is to study the existence of neutrino flavor transitions comparing the event rates at a near detector , located approximately 300 m from the pion production target , with event rate at a far detector(super kamiokande ) located at a distance of 250 km .
the event rate recorded at the far detector has been measured to be smaller that the rate extrapolated from the near detector results , with a suppression that is consistent with the @xmath117 oscillation interpretation of the atmospheric neutrino data @xcite .
the neutrino interactions collected at the near detector represent a very large sample with potentially extremely useful information .
the discussion of this data has been perhaps the highlight of the workshop .
the near detector for the k2k experiment is really a combination of three detectors .
the main flux measurement is based on a one - kiloton water detector .
this detector uses the same photo - multipliers as super
kamiokande , and has essentially the same design ( except for scale ) and the analysis for the two detectors can use the same reconstruction algorithms .
a fine grained detector sits downstream of the water detector .
it is made as a scintillating fiber tracker with water targets enclosed between layers of tracking material .
the next component of the near detector is a stack of iron plates interleaved with drift tubes to serve as a muon range detector .
both the fine grained and muon range detectors have transverse pairs of tracking planes so tracks can be reconstructed in three dimensions .
the data of the three near detectors has been discussed in a set of 4 contributions by ishida , walter , mauger and mine @xcite .
the event sample collected up to now in the near detectors amount to a very significat statistics : 27,000 events in the water detector , 8,300 in the fine grained detector and 125,000 in the iron detector .
a comparison of the data with a montecarlo simulation shows in general good agreement , demonstrating that there are no large systematic errors in the simulation chain , however small but significant effects are visible .
the fine grained detector @xcite as a whole is excellent for event classification studies since it can distinguish quasi - elastic and inelastic events .
this is performed measuring the momenta and directions of the muon and the proton in events of type : @xmath118 . in case of quasi
elastic scattering on a free nucleon , because of 4momentum conservation , the information obtained from the the measurements is redundant , and the measurements of @xmath119 , @xmath120 and @xmath121 allow to determine the initial neutrino energy and the final state proton direction . for scattering on a bound nucleus ,
since the spectator nucleons absorb some 4momentum , the predicted values represent the central values of narrow distributions .
the contribution of the inelastic channels can be separated from the quasi
elastic one because it is much broader as illustrated in fig .
[ fig : finegrained ] .
a method to discriminate between the hypothesis of @xmath21 and @xmath122 transitions is the study of the neutral current event rate .
this event rate is suppressed in the case of transitions of @xmath123 into sterile neutrinos , while it remains constant for standard @xmath21 transitions .
[ hbt ] the most important channel for the study of the neutral current event rate is the reaction : @xmath124 that has the lowest multiplicity and a clear experimental signature .
the problem is that in order to predict the event rate for this process one needs to know with sufficient precision the absolute value of the cross section , or at least the ratio with respect to a measurable charged current process .
the k2k near detector is the ideal instrument to measure this cross section ( or a cross section ratio such as @xmath125 .
the status of these measurements is described in @xcite .
approximately 5000 @xmath126 were reconstructed in a 50 ton fiducial volume in the water near detector , determining the ratio @xmath127 ) with small statistical error and an overall systematic error of order @xmath128 .
this can be applied to the 414 reconstructed @xmath126 in the atmospheric neutrino sample ( 1289 days ) of sk to obtain a result that disfavors the oscillation into sterile neutrinos .
it is well known that atmospheric neutrinos were initially considered mostly as an unpleasant background for proton decay studies , because for any detector resolution , there is a finite probability that a @xmath2 interaction will mimic a proton decay event .
it is clearly very important to estimate correctly this probability for the interpretation of the existing data , and more important for the planning of future experiments .
this probability is clearly strongly dependent on the @xmath2 interaction properties .
the exposition of the 1 kton ( 50 tons fiducial mass ) water detector to the @xmath2 fluence obtained from @xmath129 protons is roughly equivalent to a @xmath130 mton yr exposition to the atmospheric @xmath2flux and therefore allows to study experimentally the importance of the atmospheric background for @xmath32 decay studies ( see @xcite for more discussion ) .
it is necessary to estimate carefully the possible impact of systematic uncertainties in our knowledge of the neutrino cross sections for the interpretation of present and future neutrino data .
the effect of uncertatainties on @xmath26 for the interpretation of the sk atmospheric neutrino data has been addressed at this workshop in the contribution of k. kaneyuki @xcite .
this work analysed the effects of three modifications of the neutrino cross section : ( a ) a change in @xmath131 of @xmath132 , ( b ) the setting to zero of the cross section for @xmath126 coherent production on a nucleus ( the process @xmath133 @xcite ) , and ( c ) a modification of @xmath132 for the ratio @xmath134 .
the sk simulation was repeated including these modifications , and the allowed region in the plane @xmath135,@xmath136 ) was recalculated following the same standard procedure .
the allowed intervals calculated with the modified cross section are remarkably similar to the normal ones .
the effect of the @xmath137% increase of the neutral current cross section corresponds to a small ( @xmath138 ) increase in the allowed range for @xmath139 .
this can be understood noting that the neutral current events contribution to the single ring events is larger for @xmath140like events , therefore a larger neutral current cross section corresponds to a larger @xmath141 , to a smaller double ratio @xmath142 and therefore to a larger suppression of the @xmath140rate , and corresondingly to a larger @xmath139 . a @xmath132 modification of the multi pion cross section correspond to a shift of order @xmath143 in the minimum allowed value of the mixing parameter @xmath136
this can be qualitatively understood observing that in the presence of a larger contribution of the non quasi elastic component , the correlation between the muon and the neutrino becomes weaker .
the up / down asymmetry observed at the muon level is in general smaller than the true asymmetry at the neutrino level , because of finite angular resolution effects , with the dominant effect being the angle @xmath120 between the observed muon and the parent neutrino that is determined by the interaction cross section . assuming a weaker correlation , the inferred true asymmetry becomes stronger , and the mixing parameter larger .
the effect of neglecting the coherent @xmath126 cross section is found to ne negligible .
these results are very encouraging , in the sense that they demonstrate that the atmospheric neutrino result is very robust , and that the uncertainties in our knowledge of the neutrino cross section is actually very small .
this is after all not so surprising and is a consequence of the fact that essentially all effects in atmospheric neutrinos are visible in ratios such as @xmath144 or up / down , and the theoretical uncertainties tend to cancel . a problem that has attracted a fair amount of attention in the past is the question of the absolute value of the atmospheric neutrino fluxes , and the precise shape of their energy dependence .
different calculations of the atmospheric @xmath2 fluxes @xcite yield results that differ in absolute normalization by 20% or more , and also have non negligible differences in the slope of the energy spectrum .
these differences have negligible effects on the neutrino oscillation interpretation of the data , however it would be very desirable to determine the correct normalization of the fluxes , but this is of course only possible is the absolute value of @xmath26 is known with sufficient precision .
future experiments with accelerator beams aiming at more precise determination of the neutrino oscillation parameters will very likely have more stringent requirements on the control of the neutrino cross section .
a quantitative discussion of this question is still not complete .
for example , possible effects for the minos experiment have been discussed by adam para @xcite . in experiments that measure the shape of the survival probability @xmath145 as a function of the neutrino energy @xmath13 ( for a fixed pathlength @xmath10 ) it is essential to have a good reconstruction of the neutrino energy .
this also depend to a good understanding of the neutrino cross section , to correct the visible energy for the energy absorbed in the excitation of the target nucleus , and the average composition and multiplicity of the final state .
an extremely good knowledge of the neutrino cross section will be required in case the planned neutrino factories will be constructed and used for the determination of cp violating effects in neutrino flavor transitions . in this case one wants to compare the rates of processes such as : @xmath146 to and @xmath147 and one will need a control of the ratio @xmath148 to a better precision than the size of the expected asymmetries .
an interesting contribution was given by yoshihia obayashi @xcite , for the jhf collaboration .
the jhf project @xcite plans to use the 50 gev proton beam of the jaeri accelerator to produce an intense neutrino beam directed at the super kamiokande detector ( with a 295 km baseline ) .
the first phase is directed at the determination of @xmath149 studying transitions @xmath150 , while a second phase ( with a more massive 10 mt hyper kamiokande detector ) is aimed at a study of cp violating effects . a cp violation asymmetry can be defined as @xmath151 the oscillation probability can be schematically determined as : @xmath152 where @xmath153 is the observed rate of @xmath154like events , @xmath155 is the detection efficiency and @xmath156 is the expected @xmath123 fluence at the detector in the absence of oscillations .
the probability @xmath157 can be obtained with a neutrino beam of opposite sign .
it is clear that for the determination of the asymmetry parameter only the ratio @xmath158 is relevant . in fact , considering that also the determination of the expected flux depends on knowledge of the cross section , the asymmetry can be rewritten as a function of the quantity @xmath159 .
a part of the systematic uncertainties cancel in this ratio , however it is not clear if , even considering this cancellation , the present uncertainties are sufficiently small to allow the measurement without a special program of cross section measurements .
in all present and future neutrino experiments the analysis of the data will require detailed montecarlo calculations , therefore the problem of describing the neutrino interactions can be divided into two parts of similar importance : * the development of a theoretical framework to describe the interactions . * the development of montecarlo algorithms to implement this understanding . of course , since a calculation of the neutrino cross sections with hadrons is not possible from first principles , the first and most important task is to collect experimentally the necessary information .
the development of montecarlo algorithms in a non trivial problem , that has probably been given too little attention until this moment .
all the neutrino experiments that have recently collected data have used different montecarlo codes , written by members of the collaboration , and the results of the different codes have not been object of a sufficiently careful and detailed comparison .
one of the successes of the nuint01 workshop , has been in fact the discussion between different groups and the comparison between different codes .
the super kamiokande collaboration has used two different montecarlo codes , the neut code @xcite , originally developed for kamiokande , and the nuance code @xcite ( with a progenitor developed for imb ) .
the soudan2 detector has used the code neugen @xcite that is now developed to be used by minos .
the icarus collaboration has performed simulations using a code developed for ( and tuned with the data of ) the nomad detector at cern @xcite , and a second one discussed in @xcite .
the opera detector has performed its simulations with the code jetta developed for chorus , that gives special attention to the @xmath63 cross section @xcite , an additional mc for opera was presented at the workshop @xcite .
the montecarlo code of miniboone was discussed in @xcite .
this is not the place for a critical discussion of the results of this comparison , and only few comments are possible .
no major discrepancy between the different codes has been found , however several non trivial effects have been found , and in fact even non negligible bugs have already been found and corrected for .
most codes tend to use the same theoretical input namely : + ( i ) llewellyn smith @xcite for the expression of the quasi
elastic cross section on free nucleons .
+ ( ii ) smith and moniz for the implementation of the fermi gas model for quasi elastic scattering .
+ ( iii ) the resonance cross section of rein and sehgal .
+ ( iv ) the `` standard '' dis formula for high @xmath78 and @xmath1 , with one of the publically available sets of pdf s . + even if the theoretical input is the same , there are non trivial differences that are present , in particular about the implementation of the fermi gas model , the `` joining '' of the resonance production and dis scattering regimes , and the hadronization of the final hadronic state . the nuclear rescattering effects are treated ( when they are ) in significantly different ways , so effects such as pion reabsorption , pion elastic and charge
exchange scattering are very different .
the field of neutrino physics would certainly benefit significantly if some `` standard '' codes implementing correctly algorithms recognized as a possible good approximations for the problem , could become publically available , as it is the case for well established event generators in other sub fields of particle physics .
the systematic uncertainties in the neutrino interaction properties are still sufficiently large that it is probably neither possible nor really desirable to have a unique `` universal code '' , but in the era of precision studies in neutrino physics , when the requirements on the event generators will be more stringent , the development of more sophisticated and better tested instruments is clearly required .
the study of atmospheric neutrinos with energies around 1 gev has yielded remarkable results on the neutrino mass matrix .
uncertainties on the the neutrino nucleus cross section in this energy range are of order @xmath160 .
these uncertainties very likely have a small impact on the interpretation of the data , however for future studies using accelerator neutrinos in approximately the same range of energy , a more precise knowledge of the cross section would be valuable , and in fact it will be required for studies that attempt to detect and measure @xmath161 violation effects in neutrino oscillations .
the size of the uncertainties on the @xmath2 interaction properties can be very likely reduced incorporating in our description of the neutrino cross sections the understanding of nuclear effects , hadronic resonance production , and the behaviour of the nucleon structure functions at low @xmath1 that has been obtained in more recent studies in electron scattering .
the near detector of the k2k experiment has collected the largest existing sample of @xmath2 interactions in the @xmath0 gev region .
these data have a great potential to clarify several important questions on the neutrino interaction properties .
hopefully a sufficient precise determination of the normalization and energy spectrum of the @xmath2beam will allow the measurement of absolute values for the cross sections ; in any case important information about the ratios of cross sections can be obtained .
the measurement of the rate for the neutral process @xmath162 relative ( for example ) to quasi
elastic scattering , is very important to put limits on oscillations of standard neutrinos into sterile states . * a more quantitative estimate of the importance of uncertainties on @xmath26 as limiting factor for the sensitivity of future experiments on neutrino oscillations . * a complete analysis of the k2k near detector data to obtain the most of these remarkable results . * the development of a treatment of nuclear effects that incorporates modern understanding and calculation methods , going beyond the fermi gas approximation .
* more accurate modeling of pion and nucleon rescattering in nuclear matter . * an improved treatment of hadronic resonance production , with a smooth joining of this kinematical region to the deep inelastic region , using the insight allowed by the electron scattering results . * development and testing of the montecarlo algorithms that implement our knowledge of the @xmath2 interaction properties . *
last but not least : the proposal of new experimental studies aimed at the measurement of the most important / interesting properties of neutrino interactions , using the new intense @xmath2beams . see for example
the contribution of jorge morfin in these proceedings @xcite .
neutrinos can be a precious window on the physics of the unification ( beyond the standard model ) , but on the other hand they can also be an extraordinarily valuable probe on the properties of qcd and the structure of the nucleons .
the @xmath163 structure of the neutrino interactions allows to explore with an axial probe the properties of a bound system of quarks in ways that are very difficult to obtain in @xmath154scattering .
the calculation of the nucleon structure functions and their evolution in the low @xmath1 region remains an important challenge for future qcd studies . in summary the measurement of the @xmath2 and @xmath85 interaction properties
can be seen both as an instrument necessary to determine with high precision the @xmath2 masses and mixing _ and _ as a a important scientific goal per se . in the golden era of neutrino physics ,
the motivations are powerful , the difficulties challenging and the opportunities rich e varied . *
acknowledgments * i m very grateful to the organizers of nuint01 for a very interesting and frutiful workshop , and to many of the participants for instructive discussions . special thanks to maurizio lusignoli makoto sakuda , and chris walter .
|
the present uncertainties in the knowledge of the neutrino cross sections for @xmath0 gev , that is in the energy range most important for atmospheric and long baseline accelerator neutrinos , are large .
these uncertainties do not play a significant role in the interpretation of existing data , however they could become a limiting factor in future studies that aim at a complete and accurate determination of the neutrino oscillation parameters .
new data and theoretical understanding on nuclear effects and on the electromagnetic structure functions at low @xmath1 and in the resonance production region are available , and can be valuable in reducing the present systematic uncertainties .
the collaboration of physicists working in different subfields will be important to obtain the most from this available information .
it is now also possible , with the facilities developed for long baseline beams , to produce high intensity and well controlled @xmath2beams to measure the neutrino interaction properties with much better precision that what was done in the past .
several projects and ideas to fully exploit these possibilities are under active investigation .
these topics have been the object of the first @xmath2interaction ( nuint01 ) workshop .
|
in this article we consider interaction of electromagnetic field with quasi - periodic media which we define as media obtained by some smooth deformation of perfect periodic structures .
mathematically , the term smooth deformation means the biunivocal smooth mapping of the medium under consideration on the medium with perfect periodic structure .
the main goal of this article is to show that the electrodynamics of quasi - periodic media can be reduced to the electrodynamics of the perfect periodic media , and , thus , to make possible application to quasi - periodical structures of well - developed classical approaches.the correspondence between perfect and quasi - periodical media is possible when the later one is considered as the riemannian manifold with the fundamental metric tensor @xmath2 @xmath3 , where @xmath4 is the kronecker symbol and @xmath5 is the deformation tensor . to clarify this statement ,
let us remind , that in the mechanics of continuous medium there are two methods describing the deformation processes : the euler method and the lagrange method .
the co - ordinates coincide with the observer s frame of reference in the euler method .
the method of lagrange stands that the co - ordinates are mounted into a medium and are deformed together with a medium .
it should be emphasized that the co - ordinates of some point of a medium in the lagrange co - ordinate system remain unchanged during the medium deformation , whereas the co - ordinates of some point of the medium in the euler co - ordinate system are being changed . in particular , the co - ordinates of bravais lattice points of a medium with perfect periodic structure , defined before the deformation by a set of three integers @xmath6 will remain constant both during the deformation and after the deformation .
hence , the bravais lattice of a deformed ( quasi - periodic ) medium will look alike in the lagrange and the euler co - ordinate systems .
however , it is necessary to remember , that the area engaged in a quasi - periodic medium in the lagrange co - ordinate system is a riemannian manifold with the fundamental metric tensor varying from point to point .
let us give some examples how the formalism developed in this article can be used while describing the interaction of electromagnetic field with media .
generally even perfect crystals can be referred rather to the quasi - periodic media , than to the media with perfect periodic structure because there are always temperature fields of deformation at least .
the classic and quantum superlattices electrodynamics of semiconductors also can be referred to the electrodynamics of the quasi - periodic media when it is necessary to take into account the fields of the deformation , permanently existing in such structures . a big amount of articles devoted to the electrodynamics of the carbonic nanotubes is now published . in most cases
they take into the consideration the perfectly straight nanotubes with the perfect periodic structure .
actually there are always some kinds of the nanotube deformation .
it is possible to reduce the electrodynamics of such nanotubes , and furthermore the electrodynamics of ensemble of nanotubes in some matrix completely to the electrodynamics of the quasi - periodic media . in real crystals
the ensemble of the flaws , dislocations and impurities creates a field of deformations , which is usually divided on two parts .
the first one is the averaging over the ensemble and is the smooth field of deformations .
the second one describes the fluctuations of the deformations field in comparison with this smooth averaged field .
the interaction of an electromagnetic field with the crystals , that have the deformation field being the averaged smooth field of deformations , can be described by methods suggested in this article .
the last example to be considered is the x - ray diffraction optics of the elastically deformed perfect crystals .
such crystals completely correspond to the definition of quasi - periodic media given above . from the point of view of the classical electrodynamics ,
the x - ray diffraction optics is a simple version of the geometrical optics .
the only complication are coupled waves inside a crystal which appear because of the dynamic diffraction . despite the distorted bragg mirror
is one of the most important parts of almost all x - ray optics system , we have not seen papers correctly describing a dynamic x - rays diffraction on elastically deformed crystals with mean radius of curvature less or about @xmath7 whereas elastically deformed crystals with such media radiuses of curvature are of practical interest .
takagi - taupin equations @xcite-@xcite and the series of articles ( see bibliography ) , using this equations as the basic ones are valid for elastically deformed crystals with mean radius of curvature more than @xmath8 , as will be shown in part 4 .
furthermore , these papers and also paper [ 5 ] , do not show the difference between two methods describing the crystal deformation mentioned above ( euler and lagrange methods ) .
the fact is also ignored that the deformed crystal in the lagrange co - ordinate system is the riemannian manifold with a fundamental metric tensor varying from point to point in the crystal .
such disregardings lead to a paradoxical statement , that it is impossible to destinguish a perfect crystal and the deformed crystal in the experiments on x - ray beam passing through a crystal far away from the bragg condition . that contradicts both daily practice of x - ray beams application and classical electrodynamics because the deformed crystal is a nonuniform medium with refraction index and absorption coefficient depending on co - ordinates .
the first section of this article is dedicated to the electrodynamics of media with perfect periodic structure and is being auxiliary .
as the medium with periodic structure is a medium with a spatial dispersion and the relation between the vector of polarization and the electric vector is not local , the wave equation is the integro - differential one and is rather difficult to work with . in case of media with perfect periodic structure one usually proceeds to the ( @xmath9)-representation , where the integro - differential wave equation is replaced by the system of the algebraic equations .
then one derives the medium eigenmodes from the dispersion equation and gains all key results of the electrodynamics of media with perfect periodic structure .
such an algorithm defies the quasi - periodic media generalization , therefore in the first section we expound an unknown method , unwieldy and inconvenient on the first sight , that allows to substitute the integro - differential wave equation for the equivalent system of the differential equations , that describes the electromagnetic field dynamic diffraction by the media with perfect periodic structure .
this method can be naturally generalized for the case of the quasi - periodic media .
the second section is devoted to the electrodynamics of quasi - periodic media .
proceeding to the lagrange co - ordinate system allows us to apply the formalism of media with perfect periodic structure developed in the first section to the quasi - periodic media almost without changes in case we consider a quasi - periodic medium in a lagrange co - ordinate system being a riemannian manifold . in the third section
we obtain the polarization tensor of a deformed crystal in the x - ray frequency band from the first principles by the joint solution of maxwell equations and von neumann s equation for a density matrix of a crystal .
it is necessary to underline here , that the dependence of the debye - waller factor on the co - ordinates is ignored , the fourier - components of the polarization tensor of a deformed crystal coincide with fourier - components of the polarization tensor of a perfect crystal by the shape only in a lagrange co - ordinate system .
these components appear to be very complicated co - ordinate functions in the observer system ( euler co - ordinate system ) .
in the fourth section we obtain the equations system of the two - wave x - rays diffraction on the deformed crystals and after a series of simplifying approximations we show that it is possible to receive from these equations a combined takagi -taupin @xcite-@xcite equations system which appears applicable only for the deformations of the crystals with medial radius of curvature more than @xmath8 .
finally in the fifth section we use a principle of the geometrical optics locality to offer a simple formula for the bragg reflection coefficient from distorted bragg mirror and describe the x - ray quanta mirror focusing .
first we consider interaction of the electromagnetic field with a perfect periodic medium . for media with spatial dispersion ,
the wave equation for electric field vector in @xmath10-representation has the form as follows : @xmath11 where @xmath12 is the medium polarization tensor , and @xmath13 is the current density of external sources . the intrinsic property of perfect periodic media to be translationally invariant , makes it possible to expand the polarizability tensor into the fourier series : @xmath14 in the above equations @xmath15 is the bravais lattice vector , @xmath16 and @xmath17 are integer quantities , @xmath18 are the reciprocal lattice vectors , @xmath19 , @xmath20 , @xmath21 and @xmath22 are the basic vectors of the bravais and reciprocal lattices , correspondingly .
let an electromagnetic wave be incident on a periodic medium with a given bravais lattice . if the wavelength of the incident electromagnetic field is of the order of the elementary sell linear extension , solution of eq .
( 1 ) is reasonable to seek in the form of the expansion @xmath23 where the amplitudes @xmath24 vary slowly in the area with the linear dimensions less then the wavelength and @xmath25 . by substitution eqs .
( [ 2])-([3 ] ) into eq .
( [ 1 ] ) we obtain @xmath26{\bf e}_{{\bf l}}({\bf r},\omega ) + k^{2}\sum\limits_{{\bf l}^{\prime } } \int d^{3}{\bf r}^{\prime } \hat{\chi}^{{\bf l}^{\prime } } ( { \bf r}^{\prime } , \omega ) e^{-i({\bf k}+{\bf \tau } ^ { ( { \bf l})}){\bf r}^{\prime } } { \bf e}_{{\bf l+l}^{\prime } } ( { \bf r}-{\bf r } ^{\prime } , \omega ) = 0 \label{4}\ ] ] where @xmath27 and the differentiation operator @xmath28 $ ] is determined by @xmath29=\left [ \nabla + i\left ( { \bf k}+{\bf \tau } ^{({\bf l})}\right ) \right ] \times \left [ \nabla + i\left ( { \bf k}+{\bf \tau } ^{({\bf
l})}\right ) \right ] .\ ] ] the symbol @xmath30 stands for the vector product . expanding the amplitude @xmath31 into the taylor series and substituting this series into the integral in expression ( [ 4 ] ) we obtain @xmath32 @xmath33 . here and below
, the summation over repetitive indices is implicit . since @xmath34 , the estimate @xmath35^{-1}\partial \chi _ { \alpha , \beta } ^{{\bf ll^{\prime } } } /\partial k_{i}=o(1/k)$ ] holds true and , as a result , @xmath36
then , in view of that the amplitudes @xmath37 are slowly varying functions , all terms in the last expression except the first one can be omitted and , consequently , integro - differential equation ( [ 4 ] ) reduces to @xmath26{\bf e}_{{\bf l}}({\bf r},\omega ) + k^{2}\sum\limits_{{\bf l}^{\prime } } \hat{\chi}^{{\bf l}^{\prime } } ( { \bf k}+ { \bf \tau } ^{({\bf l})},\omega ) { \bf e}_{{\bf l+l}^{\prime } } ( { \bf r},\omega ) \}=0 .
\label{5}\ ] ] let @xmath38 be the radius - vector of the elementary sell center . in that case ,
any point @xmath39 inside the sell is given by @xmath40 .
taking this representation into account , let us multiply the last equation by @xmath41 and integrate the product over the elementary sell volume @xmath42 : @xmath43{\bf e}_{{\bf l}^{\prime } } ( { \bf r}_{0}+{\bf \eta } , \omega ) d^{3}{\bf \eta } \\ & & + k^{2}\sum\limits_{{\bf l}^{^{\prime \prime } } } \hat{\chi}^{{\bf l } ^{^{\prime \prime } } } ( { \bf k}+{\bf \tau } ( { \bf l}^{\prime } ) , \omega ) \int\limits_{\omega } e^{i({\bf \tau } ( { \bf l}^{\prime } ) -{\bf \tau } ^{({\bf l})}){\bf \eta } } { \bf e}_{{\bf l}^{\prime } { \bf + l}^{^{\prime \prime } } } ( { \bf r}_{0}+{\bf \eta } , \omega ) d^{3}{\bf \eta } \}=0\end{aligned}\ ] ] in the framework of the slowly varying amplitude approximation taken above , the quantities @xmath44{\bf e } _ { { \bf l}^{\prime } } ( { \bf r}_{0}+{\bf \eta } , \omega ) $ ] and@xmath45 can be treated as constants within the elementary sell and , thus , can be factored out from the integrals . in that case , since @xmath46 the equation ( [ 5 ] ) reduces to the system as follows : @xmath29{\bf e}_{{\bf l}}({\bf r } , \omega ) + k^{2}\sum\limits_{{\bf l}^{\prime } } \hat{\chi}^{{\bf l}^{\prime } } ( { \bf k}+{\bf \tau } ^{({\bf l})},\omega ) { \bf e}_{{\bf l+l}^{\prime } } ( { \bf r } , \omega ) = 0 \label{6}\ ] ] together with eq .
( [ 3 ] ) , this equation system describes the electromagnetic field dynamic diffraction by a perfect periodic medium .
when the wavelength is much larger then the medium unit cell linear size , electromagnetic fields propagating in the medium can be averaged over the elementary sell and , then , a hypotatic homogeneous medium with effective constitutive parameters can be introduced into consideration instead of the real inhomogeneous one .
polarization of this homogeneous medium is related to averaged electromagnetic field by @xmath47 , where @xmath48 is the polarization tensor which determines the linear electromagnetic response of the effective homogeneous medium . in accordance with @xcite , the effective polarizability tensor @xmath49
is expressed in terms of the fuorier transform of the microscopic polarizability tensor ( [ 2 ] ) by @xmath50 the tensor @xmath51 takes into account interaction between scattering centers and is determined by the lattice geometry : @xmath52 the function @xmath53 is an even function of co - ordinates . in centro - symmetrical lattices
it is defined by its momenta : @xmath54 the tensor @xmath51 has been evaluated analytically for some types of elementary cells @xcite . in other cases
it can be done numerically .
the basic assumption which the procedure of derivation of eqs .
( [ 6])- ( [ 8 ] ) leaned upon , is the relation ( [ 3 ] ) accounting for the property of the polarizability tensor of perfect periodic structures to be translationally invariant . under effect of deformation , this invariance , e.g. condition ( [ 3 ] ) ,
is disrupted and thus the algorithm of eqs .
( 6)-([7 ] ) derivation outlined above becomes invalid .
however , an appropriate modification based on the approach originated from the continuous medium mechanics can be applied to involve deformed periodic media into consideration . in the mechanics of continuous media
there are two alternative methods of the deformation process description , the euler method and the lagrange method . in the framework of the first one
the co - ordinate system is coupled with the viewpoint whereas the second method treats the co - ordinate system frozen into the medium and , thus , in the process of deformation the co - ordinate lines are transformed along with the medium . as a result ,
the euler co - ordinates of a given point @xmath55 are changed during the deformation whereas the lagrange co - ordinates of it , @xmath56 , remain fixed .
lagrange and euler co - ordinates are related to one another by @xmath57 where @xmath58 stand for the deformation field components . in particular , choosing the co - ordinate system @xmath59 to be related to the elementary translation vectors of the bravais lattice @xmath60 , @xmath61 , we find out the lagrange co - ordinates of the bravais lattice sites to be given by quantities @xmath62 both before and after the deformation .
thus , one can conclude that in the lagrange co - ordinates , the bravais lattice of a deformed periodic structure looks like to the bravais lattice of a perfect periodic structure and consequently , the polarizability tensor in the lagrange co - ordinates proves to be translationally invariant .
this allows us to state the relations as follows @xmath63 where @xmath64 is the reciprocal lattice covector in the lagrange co - ordinate system ; its components in the euler co - ordinates are defined by @xmath65 .
it should be emphasized that , mathematically , a deformed medium is the riemann manifold with the fundamental metric tensor as follows @xmath66 where @xmath67 is the deformation tensor of the medium : @xmath68 here @xmath69 is the metric tensor of elementary cell of the perfect periodic medium .
further mathematical analysis accounted for the properties of the riemann manifolds allows one to extend the approach being developed to media which need not be obtained by deformation of the perfect periodic medium . in order for relation (
[ 9 ] ) to hold true , a one - to - one smooth mapping ( [ 10 ] ) of the medium being considered into a perfect periodic structure is necessary and sufficient to exist .
the media for which such a mapping exists we shall refer to as quasi - periodic media . wave equation for quasi - periodic media in the lagrange co - ordinates takes the form as follows @xmath70{\bf e}({\bf \xi } , \omega ) = \frac{\omega ^{2}}{c^{2}}{\bf e}({\bf \xi } , \omega ) \nonumber \\ & & + \frac{\omega ^{2}}{c^{2}}\sum\limits_{{\bf l}}\int d^{3}{\bf \xi } ^{\prime } \sqrt{\left| g({\bf \xi } ^{\prime } ) \right| } \hat{\chi}^{{\bf l}}({\bf \xi } - { \bf \xi } ^{\prime } ; \omega ) e^{-i\tau _ { i}^{({\bf l})}\xi ^{i}}{\bf e}({\bf \xi } ^{\prime } , \omega ) + \frac{4\pi \omega i}{c^{2}}{\bf j}^{(e)}({\bf \xi } , \omega ) \label{12}\end{aligned}\ ] ] where @xmath71 is the tensor inverse to the metric tensor @xmath72 determined by eq .
( [ 11 ] ) ( @xmath73 ) , @xmath74 the procedure of derivation of basic equation describing wave diffraction by quasi - periodic media is analogous to that applied above under derivation of eqs .
( [ 6])-([7 ] ) . as in the case of perfect periodic media , solution of wave equation (
[ 12 ] ) we shell seek in the form of expansion analogous to ( [ 3 ] ) : @xmath75\xi ^{i}+ik_{i}u^{i}({\bf \xi } ) } \label{13}\ ] ] where @xmath76are the wave vector @xmath77 components in the euler co - ordinate system . as before ,
the wavelength @xmath78 is assumed to be of the order of the elementary cell linear size .
it should be emphasized that expression ( [ 13 ] ) , as different from eq .
( [ 3 ] ) , is not the expansion in terms of plane waves because the phases @xmath79 are intricate functions of co - ordinates in both euler and lagrange systems . letting in eq .
( [ 12 ] ) @xmath80 and substituting eq.([13 ] ) into ( [ 12 ] ) , one can obtain @xmath81\xi ^{i}}[-{\bf e}_{h}({\bf \xi } ) \otimes \hat{l}_{l}^{h}({\bf k}({\bf \xi } ) + { \bf \tau } ^{({\bf l } ) } , { \displaystyle{\frac{\partial } { \partial { \bf \xi } } } } ) { \bf e}^{l}({\bf \xi } ) + \hat{l}({\bf k}({\bf \xi } ) + { \bf \tau } ^{({\bf l } ) } , { \displaystyle{\frac{\partial } { \partial { \bf \xi } } } } ) + k^{2}]{\bf e}_{{\bf l}}({\bf \xi } , \omega ) \nonumber \\ & & + k^{2}\sum\limits_{{\bf l}^{\prime } } \int d^{3}{\bf \xi } ^{\prime } \sqrt{\left| g({\bf \xi -\xi } ^{\prime } ) \right| } \hat{\chi}^{{\bf l}^{\prime } } ( { \bf \xi } ^{\prime } ; \omega ) e^{i[k_{i}+\tau _ { i}^{({\bf l})}](\xi ^{i}-\xi ^{^{\prime } i})+ik_{i}u^{i}({\bf \xi -\xi } ^{\prime } ) } { \bf e}_{{\bf l+l } ^{\prime } } ( { \bf \xi -\xi } ^{\prime } , \omega ) \}=0 \label{14}\end{aligned}\ ] ] where @xmath82 are the wave vector @xmath77 components in the lagrange co - ordinate system , @xmath83 , @xmath84 and differentiation operators @xmath85 , @xmath86 are determined by the relations as follows : @xmath87\}\frac{1}{\sqrt{g}}\ { { \displaystyle{\frac{\partial } { \partial \xi ^{l } } } } + i[k_{l}({\bf \xi } ) + \tau _ { l}^{({\bf l})}]\}\sqrt{g } \nonumber \\ & & \hat{l}({\bf k}({\bf \xi } ) + { \bf \tau } ^{({\bf l } ) } , { \displaystyle{\frac{\partial } { \partial { \bf \xi } } } } ) = \frac{1}{\sqrt{g}}\ { { \displaystyle{\frac{\partial } { \partial \xi ^{l } } } } + i[k_{l}({\bf \xi } ) + \tau _ { l}^{({\bf l})}]\}\sqrt{g}g^{lj}\ { { \displaystyle{\frac{\partial } { \partial \xi ^{j } } } } + i[k_{j}({\bf \xi } ) + \tau _ { j}^{({\bf l})}]\ } \label{14a}\end{aligned}\ ] ] integral in eq .
( [ 14 ] ) can be estimated in the following way . as has been pointed out above
, the polarizability tensor @xmath88 is proportional to the amplitude of scattering of electromagnetic field by elementary cell . by this reason ,
the quantity @xmath89 quickly fall down with @xmath90 and becomes negligible at distances exceeding linear size of the elementary cell .
it means that the taylor expansion of @xmath91 in the vicinity of the point @xmath92 in the integrand of eq .
( [ 14 ] ) can be truncated beyond the first term .
further analysis allows one to reduce eq .
( [ 14 ] ) to the system as follows : @xmath93\xi ^{^{\prime } i}}{\bf e}_{{\bf l+l}^{\prime } } ( { \bf \xi -\xi } ^{\prime } , \omega ) = \nonumber \\ & & \sqrt{\left| g({\bf \xi } ) \right| } \hat{\chi}^{{\bf l}^{\prime } } ( { \bf k } ^{\prime } ( { \bf \xi } ) + { \bf \tau } ^{({\bf l})};\omega ) { \bf e}_{{\bf l+l } ^{\prime } } ( { \bf \xi } , \omega ) + i\sqrt{\left| g({\bf \xi } ) \right| } \frac { \partial } { \partial k_{i}}\hat{\chi}^{{\bf l}^{\prime } } ( { \bf k}({\bf \xi } ) + { \bf \tau } ^{({\bf l})};\omega ) \frac{\partial { \bf e}_{{\bf l+l}^{\prime } } ( { \bf \xi } , \omega ) } { \partial \xi ^{i } } \nonumber \\ & & + i\frac{\partial \sqrt{\left| g({\bf \xi } ) \right| } } { \partial \xi ^{i } } \frac{\partial } { \partial k_{i}}\hat{\chi}^{{\bf l}^{\prime } } ( { \bf k}({\bf \xi } ) + { \bf \tau } ^{({\bf l})};\omega ) { \bf e}_{{\bf l+l}^{\prime } } ( { \bf \xi } , \omega ) + \cdots \nonumber\end{aligned}\ ] ] hence the integro - differential equation ( [ 14 ] ) can be approximated with the good accurency by the differential equation @xmath94{\bf e}_{{\bf l}}({\bf \xi } , \omega ) \nonumber \\ & & + k^{2}\sum\limits_{{\bf l}^{\prime } } \sqrt{\left| g({\bf \xi } ) \right| } \hat{\chi}^{{\bf l}^{\prime } } ( { \bf k}({\bf \xi } ) + { \bf \tau } ^{({\bf l } ) } ; \omega ) { \bf e}_{{\bf l+l}^{\prime } } ( { \bf \xi } , \omega ) \}=0 \nonumber\end{aligned}\ ] ] just as for the perfect periodic media it can be proved that this equation is equivalent to the system @xmath95{\bf e}_{{\bf l}}({\bf \xi } , \omega ) + \nonumber \\ & & + k^{2}\sum\limits_{{\bf l}^{\prime } } \sqrt{\left| g({\bf \xi } ) \right| } \hat{\chi}^{{\bf l}^{\prime } } ( { \bf k}({\bf \xi } ) + { \bf \tau } ^{({\bf l } ) } ; \omega ) { \bf e}_{{\bf l+l}^{\prime } } ( { \bf \xi } , \omega ) \}=0 . \label{15}\end{aligned}\ ] ] which describes dynamical diffraction of electromagnetic field by quasi - periodical media .
differentiation operators @xmath96 , @xmath97 are determined by the formulas ( [ 14a ] ) .
if the wavelength of the incident electromagnetic field is much larger than the linear size of the lattice unit cell , the averaging procedure in a quasi - periodic medium in the lagrange co - ordinate system is identical to such procedure for perfect periodic media and leads to the equation @xmath98 with the depolarization tensor @xmath99 determined by @xmath100 where the integration is over unit cell of the perfect periodic medium and the function @xmath101 is determined by the equations ( [ 8a ] ) .
the equations ( [ 16]-[17 ] ) together with the equation @xmath70{\bf e}({\bf \xi } , \omega ) = \nonumber \\ & & \frac{\omega ^{2}}{c^{2}}{\bf e}({\bf \xi } , \omega ) + \frac{\omega ^{2 } } { c^{2}}\hat{\chi}^{avg}({\bf \xi } , { \bf k}({\bf \xi } ) ; \omega ) { \bf e}({\bf \xi } , \omega ) + \frac{4\pi \omega i}{c^{2}}{\bf j}^{(e)}({\bf \xi } , \omega ) \label{18}\end{aligned}\ ] ] are determining the electrodynamics of the quasi - periodic medium in this frequency band .
in previous section we have obtained the basic equations of electrodynamics of quasi - periodic media .
now , starting with the maxwell equations and von neumann equation for the density matrix , we derive an explicit expression for the polarizability tensor in deformed crystals , which constitute a significant class of quasi - periodic media . since the crystal unit cell is about a few angstroms in size , dynamical diffraction by crystals manifests itself in the x - ray frequency range . below we restrict consideration to this frequency range
as has been shown above , a deformed crystal in the lagrange co - ordinate system is similar to a perfect crystal in the euler system ; consequently , the lagrange co - ordinate system is preferable for evaluation of the polarizability tensor of deformed crystals .
the current density induced in the point @xmath102 by x - ray quanta in electron subsystem of the crystal can be determined by the equation @xmath103 , where @xmath104 , @xmath105 is the density matrix of the crystal , @xmath106 is the @xmath107-th component of the current density operator in electron subsystem ( summation is carried out over all crystal electrons ) . in the nonrelativistic approximation for the dirac equation ,
the current density operator of the electron @xmath108 is given by @xcite : @xmath109[\delta ( { \bf \xi } -{\bf \xi } _ { s})\frac{\hat{p}_{\mu } } { 2m}+\frac{\hat{p}_{\mu } } { 2 m } \delta ( { \bf \xi } -{\bf \xi } _ { s } ) ] \label{19}\end{aligned}\ ] ] here the value @xmath110 is the potential ( rayleigh ) part of the current operator , @xmath111 is the density operator of the @xmath112-th electron , @xmath113 is the @xmath107-th component of the electromagnetic field vector potential , @xmath114 is the sum of current operators of resonance electric and magnetic transitions , @xmath115 is the asymmetric tensor ( @xmath116 ) , @xmath117 is the momentum operator in the lagrange co - ordinates system ( @xmath118 is the covariant derivative ) , and @xmath119 @xmath120 are the pauli matrixes . in the interaction representation , the von neumann equation for the density matrix of the crystal interacting with x - ray quanta has the form as follows : @xmath121 , \nonumber \\ \hat{h}_{e\gamma } ^{\prime } ( t ) & = & e^{\frac{i}{\hbar } \hat{h}_{0}t}\hat{h }
_ { e\gamma } ( t)e^{-\frac{i}{\hbar } \hat{h}_{0}t}. \label{20}\end{aligned}\ ] ] here @xmath122 stands for the unperturbed hamiltonian of the crystal while @xmath123 describes its interaction with x - ray quanta , @xmath124 accordingly to the perturbation theory the von neumann equation solution can be represented by the series @xmath125 , where @xmath126is proportional to the @xmath127-th power of the electromagnetic field : @xmath128 @xmath129 . substituting this series into eq .
( [ 20 ] ) and equating the terms of the same power in@xmath130 , we obtain @xmath131.\ ] ] note that @xmath132 and @xmath133 in the absence of x - ray quanta , i.e. , @xmath134 @xmath135 , where @xmath135 is the unperturbed crystal density matrix .
further @xmath135 is assumed to be given by the equilibrium ( gibbs ) density matrix , @xmath136 where @xmath137 is the free energy and @xmath138 is the crystal temperature .
the average value of the current density operator in the interaction representation can be obtained from the formula @xmath139 where @xmath140 is the current operator in the heisenberg representation : @xmath141,\cdots , \hat{h}_{e\gamma } ^{^{\prime } } ( t_{n } ) ] .
\nonumber\end{aligned}\ ] ] here the time moments are put in the order : @xmath142 . in the linear approximation with respect to electromagnetic field
, this equation reduces to : @xmath143 \nonumber \\ & & -sp\hat{\jmath}_{\mu } ( { \bf \xi } ^{\prime } -{\bf \xi } _ { s^{^{\prime } } } ) e^{\frac{i}{\hbar } \hat{h}_{0}(t - t_{1})}\hat{\jmath}^{\beta } ( { \bf \xi } -{\bf \xi } _ { s})e^{-\frac{i}{\hbar } \hat{h}_{0}(t - t_{1})}\hat{\rho}_{0}]\}a^{\mu } ( { \bf \xi } ^{\prime } , t_{1 } ) , \nonumber\end{aligned}\ ] ] where the summation is carried out over all crystal electrons .
let us present co - ordinates of the @xmath112-th electron in the following way : @xmath144 where @xmath145 stands for the co - ordinates of the unit cell which the @xmath112-th electron belongs to , @xmath146 are the co - ordinates of the @xmath147-th atom inside this unit cell .
in such situation , the identity @xmath148 holds true and , thus , the density of current induced by the x - ray quanta in the crystal is represented by @xmath149 \nonumber \\ & & -sp[\hat{\jmath}_{l,\mu } ( { \bf \xi } ^{\prime } -{\bf \xi } _ { { \bf n}}-{\bf \xi } _ { l}-{\bf \eta } ( { \bf n},l))e^{\frac{i}{\hbar } \hat{h}_{0}t^{^{\prime } } } \hat{\jmath}_{l}^{\beta } ( { \bf \xi } -{\bf \xi } _ { { \bf n}}-{\bf \xi } _ { l}- { \bf \eta } ( { \bf n},l))e^{-\frac{i}{\hbar } \hat{h}_{0}t^{\prime } } \hat{\rho } _ { 0}]\}a^{\mu } ( { \bf \xi } ^{\prime } , t - t^{\prime } ) .
\nonumber\end{aligned}\ ] ] here @xmath150 is the density operator of the electrons in @xmath147-th atom and @xmath151 is the current density operator in @xmath147-th atom . in this equation
we have neglected the components @xmath152\hat{\rho}_{0}\right\ } , { \bf n\neq n}^{\prime } $ ] or @xmath153 describing the nonradiative migration of excitation from @xmath147-th atom to @xmath154-th one . in another words , we have neglected delocalization of exciton in crystal . besides
, we leave out the peculiarities of x - ray quanta interaction with the band electrons , assuming all the electrons being localized on the atoms .
these approximations are proved to be correct for situation when the incident photon energy exceeds significantly the energy of any resonance transition in the crystal .
indeed , this condition allows us to restrict ourselves to the impulse approximation @xcite where there is no difference between energy spectra of band electrons and electrons localized nearby the nuclei : the electrons density periodicity in the crystal is of the only importance .
that is why , describing the interaction of x - ray quanta with the crystal , we use the terminology appropriate rather to the isolated periodically arranged atoms devoid of zone structure then to the crystal .
it should be noted that the above assumptions are not applicable to the x - ray spectroscopy where both electron delocalization and peculiarities of the x - ray quanta interaction with band electrons should be taken into consideration .
however , this problem is beyond the scope of this article and will be considered elsewhere .
the remarks made above allow us to present the crystal hamiltonian @xmath122 by @xmath155 , where @xmath156 is the hamiltonian of the phonon subsystem and @xmath157 is the hamiltonian of the @xmath147-th atom in the unit cell . for the atomic ground state in the crystal
it is possible to neglect the influence on this ground state of elementary crystal excitations ( including influence of interaction with phonons ) .
in such a case , the crystal density matrix @xmath135 can be presented by the product @xmath158 of electron subsystem density matrix @xmath159 and phonon subsystem density matrix @xmath160 , where @xmath161 , and @xmath162 are the free electron and phonon energies . as different from the ground state , the contribution of crystal elementary excitations can not be neglected for excited atoms because they lead to the appearance of the new decay channels of the exited state and , consequently , increase the @xmath163-width of the exited state .
therefore , we will assume the hamiltonians of ground and exited states to be different .
now , let us discuss how the phonon spectrum is modified under effect of the crystal deformation . in the harmonic approximation , the phonon subsystem hamiltonian of the deformed crystal in the lagrange co - ordinate system takes the form as follows : @xmath164 where @xmath165 is the @xmath166-th component of the momentum covector of the @xmath147 -th atom in the @xmath167-th cell presented in the lagrange basis .
quantum - mechanically , the momentum is given by @xmath168 with @xmath169 as the covariant derivative .
further we assume the deformation tensor to be constant over the unit cell ; then , @xmath170 .
this approximation allows us to reduce the hamilton canonical equations to @xmath171 where @xmath172 and @xmath173 is the dynamical matrix of the crystal given in the lagrange co - ordinates . the above equation for @xmath174 can be rewritten as @xmath175 where the deformation tensor @xmath176 is considered as a parameter .
then the translation invariance of hamiltonian ( [ 21 ] ) allows us to write down @xmath177 ; this leads us to the equation @xmath178 where @xmath179 is the crystal lattice dynamical matrix in the fourier representation .
hence , we immediately obtain the dispersion equation @xmath180 as the bravais lattice of the deformed crystal in the lagrange basis coincides with the bravais lattice of the perfect crystal in the euler co - ordinates system , the matrix @xmath181 coincides with the dynamical matrix of the perfect crystal .
consequently , dispersion equation ( [ 22 ] ) does not contain undefined quantities .
since the matrix @xmath182 is hermitian , the equation roots @xmath183 are real quantities ; here @xmath184 with @xmath185 as the number of atoms in the crystal cell .
this matrix is hermitian , its eigenvectors corresponding to these roots satisfy the conditions of orthonormalization and fullness @xmath186 and allow us to implement the normal modes of the crystal @xmath187 @xmath188 @xmath189 . for these modes we have @xmath190 where @xmath191 and @xmath192 is the vector of the crystal reciprocal lattice .
next , using the standard method we obtain the hamiltonian of the crystal phonon subsystem in the harmonic approximation @xmath193 \label{23}\ ] ] where @xmath194 and @xmath195 are the phonon creation and annihilation operators .
it is easy to verify that the proposed algorithm of accounting the influence of deformation on the crystal phonon spectrum also suits the case of anharmonic oscillations of the crystal lattice . at the end
we will make two remarks .
the first : since the deformation tensor @xmath196 is the essence of the smooth co - ordinates function , the equations ( [ 22])-([23 ] ) are local .
the discussed above method of accounting of deformation influence on the crystal phonon spectrum is the approximation , that can be considered precise enough only when the length of the phonon free path in the crystal is less then the linear dimensions of the area where the deformation tensor variation can be observed .
the second remark concerns the fact that the introduced parametric method of accounting the deformation influence on the crystal phonon spectrum is not something especially new in the solid - state physics : this method is being the standard one in accounting the deformation influence on the energy spectrum of band electrons ( see for example @xcite ) .
after the discussion made above let us continue the calculations of the current in the deformed crystals that was induced by x - ray quanta .
first let us take up the rayleigh component of the current @xmath197 that makes the main contribution . as @xmath198 and @xmath199 in the coulomb calibration ,
using the fourier series @xmath200 , we obtain the formula for rayleigh component of the current @xmath201 the value @xmath202 is the essence of debye - waller factor where @xmath203 is to be determined by the equation @xmath204 accordingly the debye - waller factor in the deformed crystals is the function of the deformation tensor and consequently is the function of co - ordinates determined by the equations ( [ 22])-([23 ] ) . as @xmath205 is the form factor of the @xmath147-th atom unit cell , the standard formula @xmath206 , where @xmath207 is the number of the unit cells in the volume unit , allows us to obtain the final formula for the rayleigh component of the current @xmath208 or the formula for the rayleigh component of the deformed crystal polarization tensor in the x - ray frequency range @xmath209 in the frequency range where the x - ray quanta energy is greater then the binding energy of any electron in the crystal , the current component describing the absorption of quanta by the atoms and the consequent decay of the exited state which atom passed into by means of quantum absorption is usually much smaller then the rayleigh component ( in the x - ray optics it is called the dispersion correction ) .
nevertheless the accounting of this component is very important as it determines the absorption of x - ray quanta in this crystal . using the fourier decomposition @xmath210 ,
this current component can be written down in the following way @xmath211 \nonumber \\ & & -sp[\hat{\jmath}_{l,\mu } ( { \bf k}^{\prime } ) e^{-ik_{i}^{\prime } \eta ^{i } ( { \bf n},l)}e^{\frac{i}{\hbar } ( \hat{h}_{l}+\hat{h}_{ph}({\bf \xi ) } ) t^{\prime } } e^{-ik_{i}\eta ^{i}({\bf n},l)}\hat{\jmath}_{l}^{\beta } ( { \bf k})e^{-\frac{i}{\hbar } ( \hat{h}_{l}+\hat{h}_{ph}({\bf \xi } ) ) t^{\prime } } \rho _ { 0}^{(e)}(l)\hat{\rho}_{0}^{(ph)}({\bf \xi } ) ] \}e^{i\omega t^{\prime } } e^{\mu } ( { \bf \xi } ^{\prime } , \omega ) \nonumber\end{aligned}\ ] ] in the x - ray frequency range under the consideration @xmath163 ( the width of the exited atom state in the crystal ) is about @xmath212 , that is why in the x - ray diffraction optics the assumption @xmath213 @xmath157 is being considered .
this assumption means that we neglect the influence of raman scattering of x - ray quanta on the phonons .
if we take this assumption and use the standard formulas @xmath214 and @xmath215 , we obtain for @xmath216 the following equation @xmath217e^{\mu } ( { \bf \xi } ^{\prime } , \omega ) \nonumber\end{aligned}\ ] ] where @xmath218 is the @xmath147-th atom ground state vector , @xmath219 is the static weight of the ground state with the full set of quantum numbers @xmath220 except for the angular momentum , @xmath221 is the energy of this state , @xmath222 is the total angular momentum of the atom , @xmath223 is the projection of the total angular momentum on the quantization axis .
having used the formalism developed in @xcite for the description of decay and the lifetime of virtual states of quantum systems in the case , when the spectrum of the hamiltonian of the system is continuous , we obtain the following formula @xmath224 here @xmath225 and @xmath226 are the densities of the exited atom states with energies @xmath227 and @xmath228 accordingly , @xmath229 is the matrix diagonal element of the level shift operator determined by the equation @xcite @xmath230 where @xmath231 is the hamiltonian of the interaction of the @xmath147-th atom in the exited state with the crystal , @xmath232 is the operator of the projection on the state @xmath233 .
the imaginary part of matrix element @xmath234 is determined by formula @xmath235 where the summation is taken over all decay channels of the excited atom state and is the essence of @xmath163 ( the width of the crystal atom excited state ) .
the term in the first part of formula ( [ 26 ] ) is proportional to @xmath236 and takes into account the interaction of the atom in the excited state with the crystal and in particular allows to account such an effect as exafs in the crystal polarization tensor . to describe this interaction
it is enough to let @xmath237 , where @xmath238 is the watson pseudo - potential determined by the equation @xcite : @xmath239 where @xmath240 is the matrix of scattering by @xmath154 -th crystal atom , @xmath241 is the crystal ground state vector , @xmath242 is the operator of the projection on the crystal ground state , @xmath243 is the propagator ( for details look at @xcite ) . here
we will restrict ourself only with the first term of ( [ 26])-th right part since the description of the contribution of interaction of exited atom with the crystal to the crystal polarization tensor is worth writing a separate article .
for the further calculations we will use the atom current density operator multipole decomposition @xcite,@xcite @xmath244 where @xmath245 is the multipolarity , @xmath246 @xmath247 ; @xmath248 and @xmath249 are the operators of the electric , magnetic and charge momenta of a multipole atom ; @xmath250 are the spherical vectors of electric , magnetic and charge type @xcite,@xcite , @xmath251 is the unit covector in the covector @xmath77 direction . since the field of x - ray quanta in the crystal with a high precision can be treated as transversal one the components with @xmath249 in this decomposition can be neglected right away .
as the magnet multipole transitions in atoms are much smaller then the electric multipole transitions we will content our self only with the electric multipoles . having used the spherical vectors @xmath252 properties , wigner - eckart and ziggert theorems , we obtain the equation @xmath253 where @xmath254 and @xmath255 are the reduced matrix elements of the multipole charge momenta of the crystal unit cell @xmath147-th atom ; @xmath256 is the unit covector along covector @xmath257 , @xmath258 are these covectors lengths ; @xmath259 is the polarization matrix with its elements in spiral bases @xmath260 , where @xmath261 and @xmath262 are the orthonormalized covectors in the planes orthogonal to @xmath263 and @xmath264 accordingly are expressed through the wigner functions @xmath265 if to choose the axis @xmath266 in the covectors @xmath77 and @xmath267 plane and to mark these covectors angles relative to the axis @xmath266 as @xmath268 and @xmath269 the formulas for the elements of the polarization matrix are sufficiently simplified : @xmath270 by substituting the formulas ( [ 26])-([27a ] ) to the expression for @xmath271 , we obtain @xmath272 then the polarization tensor of the deformed crystal in the x - ray frequency range is determined by the equation @xmath273 let us discuss the obtained expression .
first of all let us note that in ( [ 29 ] ) the covector @xmath77 and the frequency @xmath10 are not bound together by any equation . as it is seen from ( [ 25 ] ) , ( [ 28 ] ) and ( 29 ) , the crystal polarizability in the x - ray frequency range is a scalar value and the relation between the current induced in the crystal by x - ray quanta and electric vector or between polarization vector and electric vector is local only if we neglect the so - called dispersion correction and , consequently , we neglect the crystal absorption of x - ray quanta .
if we want to account the absorption in the crystal , the statements like the polarization vector in the x - ray frequency range is related to the electric vector by the equation @xmath274 , where @xmath275 is the crystal polarizability , and is being a scalar periodic function are just incorrect .
when we take into account the dispersion correction , we must understand that the crystal polarizability in the x - ray frequency range is in principle the tensor but not the scalar value and the relation between the polarization vector and the electric vector is unlocal .
that is why the procedure of derivation of the equations ( [ 6 ] ) and ( [ 15 ] ) in parts @xmath276 and @xmath277 ( they describe the dynamic diffraction of the electromagnetic fields on the media with perfect periodic and quasi - periodic structures ) accounts the nonlocality of the response of the medium to the electromagnetic field and is also true for the x - ray quanta diffraction on the deformed crystals .
the polarizability tensor of the medium is not being a member of the equations ( [ 6 ] ) and ( [ 15 ] ) but his fourier - components @xmath278 , that have the relation between the wave covector @xmath77 and the frequency @xmath10already defined : @xmath279 . if to note that electromagnetic fields with wave covectors @xmath77 and @xmath280 interact effectively only in vicinity to the bragg conditions @xmath281 , one can easily obtain the following formula for @xmath278 from ( [ 29 ] ) : @xmath282 it is worth to note that for @xmath283 the condition @xmath284 is true , and as @xmath285 , then @xmath286 is definitely a scalar value . for the wave covectors @xmath77 and @xmath287 ( with magnitudes in vicinity ) * * near the bragg condition the @xmath288 condition takes place , where @xmath289 is the angle width of the bragg table , i.e. is the value about @xmath290 radians .
consequently for the wave covectors @xmath77 and @xmath287 near the bragg condition we can let @xmath291 with high precision and can consider @xmath292 being the scalar value @xmath293 .
then for @xmath294 we have the final formula @xmath295\ } \label{30}\end{aligned}\ ] ] with @xmath296 determined by the expression @xmath297 and being nothing but the full cross - section of the photo - ionization ( photo - absorption ) of the @xmath147-th atom in the crystal unit cell and @xmath298 is the differential photo - ionization cross - section .
these values are presently rather well analyzed both theoretically and experimentally [ 15 ] .
it is worth to underline that while deriving formula ( [ 30 ] ) we did nt use the condition of the long - wave assumption @xmath299 with @xmath300being the atom linear size since this condition for the x - ray radiation with the wavelength about @xmath301 is not true .
in this part we will discuss the x - ray quanta diffraction on the elastically deformed crystals . as the x - ray quanta field in crystals
can be considered as a transverse field with high accuracy , the dynamic diffraction on the deformed crystals in the two - wave approximation according to the formulas ( [ 14a ] ) , ( [ 15 ] ) and ( [ 30 ] ) is described by the following equations : @xmath302\xi ^{i}+ik_{i}u^{i}({\bf \xi } ) } ; \nonumber \\ & & \left [ \frac{1}{k^{2}}\hat{l}({\bf k}({\bf \xi } ) , { \displaystyle{\frac{\partial } { \partial { \bf \xi } } } } ) + 1\right ] { \bf e}_{{\bf 0}}({\bf \xi } , \omega ) + \sqrt{\left| g({\bf \xi } ) \right| } \chi _ { ( { \bf 0})}({\bf \xi } , \omega ) { \bf e}_{{\bf 0}}({\bf \xi } , \omega ) + \sqrt{\left| g({\bf \xi } ) \right| } \chi _ { ( { \bf l})}({\bf \xi } , \omega ) { \bf e}_{{\bf l}}({\bf \xi } , \omega ) = 0 \nonumber \\ & & \left [ \frac{1}{k^{2}}\hat{l}({\bf k}({\bf \xi } ) + { \bf \tau } ^{({\bf l } ) } , { \displaystyle{\frac{\partial } { \partial { \bf \xi } } } } ) + 1\right ] { \bf e}_{{\bf l}}({\bf \xi } , \omega ) + \sqrt{\left| g({\bf \xi } ) \right| } \chi _ { ( { \bf 0})}({\bf \xi } , \omega ) { \bf e}_{{\bf l}}({\bf \xi } , \omega ) \nonumber \\ & & + \sqrt{\left| g({\bf \xi } ) \right| } \chi _ { ( -{\bf l})}({\bf \xi } , \omega ) { \bf e}_{{\bf 0}}({\bf \xi } , \omega ) = 0 \label{31}\end{aligned}\ ] ] by definition for the elastically deformed crystals the condition @xmath303 holds true . then owing to the fact that @xmath304
, we can assume in ( [ 31 ] ) @xmath305 , @xmath306 .
because the deformation field and accordingly the metric tensor @xmath307 can be considered to be the constant values in the area with linear dimensions about the x - ray quanta wavelength , the constant terms containing @xmath308 can be neglected .
the terms @xmath309 in the assumption of slowly varying amplitudes can also be neglected .
if one introduces the designations @xmath310 for the guiding cosines of wave covector @xmath77 of x - ray quanta incident on the perfect crystal and @xmath311 for the guiding cosines of wave covector @xmath77 of x - ray quanta diffracted on the perfect crystal , then by accounting the assumptions made above and the fact that in the lagrange co - ordinate system there takes place the identity @xmath312whis @xmath313 as the wave covector @xmath77 guiding cosines in the lagrange co - ordinate system , the equations system ( [ 31 ] ) can be sufficiently simplified @xmath314 { \bf e}_{{\bf 0}}({\bf \xi } , \omega ) + \chi _ { ( { \bf l})}({\bf \xi } , \omega ) { \bf e}_{{\bf l}}({\bf \xi } , \omega ) = 0 ; \nonumber \\ & & \{2i\frac{1}{k}[\sqrt{1+\alpha } n_{i}^{({\bf l})}+n_{\mu } ^{({\bf 0})}\frac{\partial u^{\mu } ( { \bf \xi } ) } { \partial \xi ^{i}}]g^{ij } { \displaystyle{\frac{\partial } { \partial \xi ^{j } } } } + \chi _ { ( { \bf 0})}({\bf \xi } , \omega ) + 1 \nonumber \\ & & -[\sqrt{1+\alpha } n_{i}^{({\bf l})}+n_{\mu } ^{({\bf 0})}\frac{\partial -u^{\mu } ( { \bf \xi } ) } { \partial \xi ^{i}}]g^{lj}[\sqrt{1+\alpha } n_{j}^ { ( { \bf l})}+n_{\mu } ^{({\bf 0})}\frac{\partial u^{\mu } ( { \bf \xi } ) } { \partial \xi ^{j}}]\}{\bf e}_{{\bf l}}({\bf \xi } , \omega ) \nonumber \\ & & + \chi _ { ( -{\bf l})}({\bf \xi } , \omega ) { \bf e}_{{\bf 0}}({\bf \xi } , \omega ) = 0 . \label{32}\end{aligned}\ ] ] this equations system describes two - wave dynamic diffraction of x - ray quanta on the deformed crystals and is the system that will be considered as basic one .
there is takagi -taupin equations system that is usually used in the x - ray optics works .
we ll show that the takagi -taupin equations system can be obtained from the system ( [ 32 ] ) after a set of simplifying assumptions .
let us make the following assumptions : 1 ) @xmath315 , @xmath316 ( we recall the metric tensor formula @xmath317 , i.e. we neglect all the values of @xmath318 order ) ; \2 ) @xmath319,@xmath320g^{ij } { \displaystyle{\frac{\partial } { \partial \xi ^{j } } } } { \bf e}_{{\bf l}}({\bf \xi } , \omega ) \thickapprox n_{i}^{({\bf l})}\delta ^{ij } { \displaystyle{\frac{\partial } { \partial \xi ^{j } } } } { \bf e}_{{\bf l}}({\bf \xi } , \omega ) $ ] ; \3 ) @xmath321 , i.e. we neglect the dependence of debye - waller factor on the co - ordinates . placing these approximated equations into the system of equations ( [ 32 ] ) , we obtain for the slowly varying wave amplitudes @xmath322 { \bf e}_{{\bf 0}}({\bf \xi } , \omega ) + \chi _ { ( { \bf l})}(\omega ) { \bf e}_{{\bf l}}({\bf \xi } , \omega ) = 0 ; \nonumber \\ & & \left [ 2i\frac{1}{k}n_{i}^{({\bf l})}\delta ^{ij } { \displaystyle{\frac{\partial } { \partial \xi ^{j } } } } + \frac{1}{k}\sqrt{1+\alpha } n_{i}^{({\bf l})}\delta ^{ij}\tau _ { \mu } { \displaystyle{\frac{\partial u^{\nu } ( { \bf \xi } ) } { \partial \xi ^{j } } } } -\alpha + \chi _ { ( { \bf 0})}(\omega ) \right ] { \bf e}_{{\bf l}}({\bf \xi } , \omega ) \nonumber \\ & & + \chi _ { ( -{\bf l})}(\omega ) { \bf e}_{{\bf 0}}({\bf \xi } , \omega ) = 0 . \nonumber\end{aligned}\ ] ] next , let us introduce vectors of @xmath323 and @xmath324 polarization @xmath325 being standard for the x - ray optics ( it is taken into account that we are applying the lagrange co - ordinates system and polarization vectors are the co - ordinate functions ) . the values @xmath326 and @xmath327 , where @xmath328 is the kristoffel symbol , that appear during the polarization vectors differentiation and can be neglected since the metric tensor remains a constant value in the area with linear dimensions about the x - ray quanta wavelength .
then if we assume @xmath329 in the second equation , these equations look as follows @xmath322 e_{{\bf 0}}({\bf \xi } , \omega ) + c({\bf \xi } ) \chi _ { ( { \bf l})}(\omega ) e_{{\bf l}}({\bf \xi } , \omega ) = 0 ; \nonumber \\ & & \left [ 2i\frac{1}{k}n_{i}^{({\bf l})}\delta ^{ij } { \displaystyle{\frac{\partial } { \partial \xi ^{j } } } } + \frac{1}{k}n_{i}^{({\bf l})}\delta ^{ij}\tau _ { \mu } { \displaystyle{\frac{\partial u^{\nu } ( { \bf \xi } ) } { \partial \xi ^{j } } } } -\alpha + \chi _ { ( { \bf 0})}(\omega ) \right ] e_{{\bf l}}({\bf \xi } , \omega ) \nonumber \\ & & + c({\bf \xi } ) \chi _ { ( -{\bf l})}(\omega ) e_{{\bf 0}}({\bf \xi } , \omega ) = 0 \label{33}\end{aligned}\ ] ] where @xmath330 is the polarization factor : @xmath331 for @xmath324polarization . for @xmath323 polarization
we have a bulky formula @xmath332 as @xmath333 , we can assume @xmath334 i.e. we can assume @xmath335 to be the same as that for the perfect crystal . in this case the equations ( [ 33 ] ) for the slowly changing amplitudes @xmath336 fully coincide with takagi -taupin equations .
it is worth mentioning that in works mentioned above , in the formula for the x - ray quanta field in the crystal @xmath337 the term @xmath338 in the expression for the passed and diffracted wave phases is left out for some reason and it is the expression that determines the x - ray quanta focusing by the distorted crystal both in laue and in bragg geometries . to determine the application field of simplifying assumptions made above and accordingly to determine the takagi -taupin equations application field let us determine the plane - parallel plate with thickness @xmath339
let this plate to be deformed by the bending momentums @xmath340 and @xmath341 uniformly distributed over the plate sides .
then the deformation field can be introduced by the following way @xcite : @xmath342 \nonumber \\ & & -\frac{1}{\rho _ { 2}}\left [ b_{25}(\xi ^{3}+\frac{h}{2})^{2}+b_{26}\xi ^{2}(\xi ^{3}+\frac{h}{2})+2b_{12}\xi ^{1}(\xi ^{3}+\frac{h}{2})\right ] ; \nonumber \\
u^{2}({\bf \xi } ) & = & -\frac{1}{\rho _ { 1}}\left [ b_{14}(\xi ^{3}+\frac{h}{2})^{2}+2b_{12}\xi ^{2}(\xi ^{3}+\frac{h}{2})+2b_{16}\xi ^{1}(\xi ^{3}+\frac{h}{2})\right ] \nonumber \\ & & -\frac{1}{\rho _ { 2}}\left [ b_{24}(\xi ^{3}+\frac{h}{2})^{2}+2b_{22}\xi ^{2}(\xi ^{3}+\frac{h}{2})+b_{26}\xi ^{1}(\xi ^{3}+\frac{h}{2})\right ] ; \nonumber \\ u^{3}({\bf \xi } ) & = & \frac{1}{\rho _ { 1}}\left [ b_{11}(\xi ^{1})^{2}+b_{12}(\xi ^{2})^{2}-b_{13}(\xi ^{3}+\frac{h}{2})^{2}+b_{16}\xi ^{1}\xi ^{2}\right ] \nonumber \\ & & + \frac{1}{\rho _ { 2}}\left [ b_{12}(\xi ^{1})^{2}+b_{22}(\xi ^{2})^{2}-b_{23}(\xi ^{3}+\frac{h}{2})^{2}+b_{26}\xi ^{1}\xi ^{2}\right ] ; \label{34}\end{aligned}\ ] ] where @xmath343 @xmath344 @xmath345 @xmath346 ( @xmath347 ) is the crystal elastic constants .
the fig.1 shows this deformed silicon plate for the parameters @xmath348 and @xmath349 and the thickness @xmath350 , and fig.2 shows the * * average radius of curvature of reflecting external surface of this plate . as fig.2 shows , the bent silicon plate average radius of curvature with such choice of parameters depends on the co - ordinates of a point on the reflecting surface and is being the value about @xmath351 .
let us discuss the most simple situation : the absorption of x - ray quanta passing through the deformed silicon plate ( bugger law).for this situation * f*rom equations ( [ 32 ] * ) * we obtain ( from equations ( [ 32 ] ) ) @xmath352 { \bf e}_{{\bf 0}}({\bf \xi } , \omega ) = 0,\ ] ] for the simplicity we will neglect the dependence on debye - waller factor on the co - ordinates and from takagi -taupin equations ( [ 33 ] ) @xmath353 e_{{\bf 0}}({\bf \xi } , \omega ) = 0.\]]then solving the first equation by the characteristics method , i.e. by solving the system of the ordinary differential equations @xmath354 with the initial conditions @xmath355 @xmath356 , where @xmath357 are the entrance co - ordinates of the x - ray beam incident on the crystal , we obtain the following equation for transition function : @xmath358where @xmath339 is the perfect crystal length . the x - ray quanta optical path @xmath359 in the deformed crystal can be obtained by solving the equation @xmath360 , where @xmath361 is the solution of the ordinary differential equations system mentioned above . for the second equation ,
the solution i.e. the gating function can be easily obtained and does not differ from the perfect crystal plane - parallel plate at all .
@xmath362where @xmath363 is the angle between the x - ray quanta incident on the crystal and the normal to the perfect crystal entrance plane directed inside the crystal .
fig.3 show these transition functions for @xmath364 @xmath365 and quanta energies @xmath366 . at the upper surface of this figure
is presented the transition function for the plane - parallel perfect silicon crystal plate and the lower surface describes the x - ray quanta absorption by the deformed crystal .
the distinction shows itself in the term of the second order and is very sufficient for such a thin plate .
it means that even in such a simple situation as the absorption of x - ray quanta passing through the deformed crystal , the takagi -taupin equations do nt correctly describe the real experimental situation ( this distinction can be neglected only for the crystal deformation with the average radius of curvature more then @xmath8 ) .
to finally define the application range of takagi -taupin equations let us compare two terms : \1 ) @xmath367g^{lj}[\sqrt{1+\alpha } n_{j}^{({\bf l})}+n_{\mu } ^{({\bf 0})}\frac{\partial u^{\mu } ( { \bf \xi } ) } { \partial \xi ^{j}}]$ ] in the second equation of the equation system ( [ 32 ] ) and \2 ) @xmath368 in the second equation of the takagi -taupin equations system ( [ 33 ] ) .
figure 4 shows the difference @xmath369g^{lj}[\sqrt{1+\alpha } n_{j}^{({\bf l})}+n_{\mu } ^{({\bf 0})}\frac{\partial u^{\mu } ( { \bf \xi } ) } { \partial \xi
^{j}}]-\frac{1}{k}n_{i}^{({\bf l})}\delta ^{ij}\tau _ { \mu } { \displaystyle{\frac{\partial u^{\nu } ( { \bf \xi } ) } { \partial \xi ^{j}}}}\]]between these values for the deformed silicon plate shown on the fig.4 - 6 for the energies @xmath370 and @xmath366 of the quanta .
if one notify that for energies @xmath370 and @xmath366 we have @xmath371,@xmath372 and @xmath373 accordingly , this figure shows that by accepting the simplifying assumptions in the equations ( [ 32 ] ) leading to the takagi -taupin equations we reject the values that are 10 times greater than the values we keep ( for the crystals with the average radius of curvature less than @xmath7 ) while obtaining these equations .
this difference shown in fig.4 becomes less then @xmath374 for the deformations with the curvature average radius more then @xmath8 , i.e. very slightly deformed crystals .
that is why one should use exactly the equations system ( [ 32 ] ) for the theoretical analysis .
this system is a little bit more complicated but is suitable for the description of the x - ray optics devices being met with in practice .
the equations ( [ 32 ] ) ( equations for the slowly varying amplitudes of the field of x - ray quanta in the two - wave dynamic diffraction on the deformed crystals ) for the arbitrary deformation field can be solved in general only numerically . in the article we underlined more than once that the x - ray optics can be considered as the geometrical optics from the point of view of general electrodynamics .
that is why to obtain the coefficient of bragg reflection from the distorted bragg mirror we will use the geometrical optics locality principle ( see for ex .
@xcite ) that evidently was first time introduced in electrodynamics by v.a .
fock @xcite who investigated the reflection of the electromagnetic wave with smooth arbitrary shape front on the smooth arbitrary shaped surface .
according to the locality principle the wave reflection in the every mirror point acts like if the incident wave is being smooth one and the curvilinear surface in this point is replaced by the tangent plane .
but it is necessary to take into account that the incident and the reflection angle change their values from point to point on this surface .
consequently the locality principle allows us to use the reflection coefficients obtained for the plane waves and plane surfaces , and in particular the bragg reflection coefficient for the diffraction x - ray optics as the primary approximation .
the bragg reflection coefficient in the case of the plane wave reflecting from the parallel - sided perfect crystal thick plate is determined by formula : @xmath375 where @xmath376 is the polarization factor mentioned above , @xmath377 is the asymmetry parameter , @xmath378 , @xmath379 is the parameter of precise bragg condition deviation .
let the x - ray quanta wave packet incident on the bragg mirror with arbitrary smooth deformation field @xmath380 has the electric vector for quanta with frequency @xmath10 ( or the energy @xmath381 ) determined in the lagrange co - ordinates system by the equation @xmath382 where @xmath383 are wave vector guiding cosines and @xmath384 and @xmath385 are the slowly changing amplitude and phase of the x - ray quanta incident on the mirror . then according to the locality principle of geometrical optics this reflection coefficient of the wave packet for distorted bragg mirror has the functional form like that for the plane wave and plane mirror .
but it is necessary to take into account that the values @xmath386 and @xmath166 for the distorted bragg mirror change from point to point on this mirror surface . if the dependence @xmath387 on co - ordinates is determined by formula ( [ 30 ] ) , the dependencies of @xmath376,@xmath107 and @xmath166 on the co - ordinates can be defined by the following ideas .
the wave covector components of the x - quanta incident on the distorted bragg mirror in the lagrange co - ordinate system to be determined by the equation @xmath388 .\ ] ] then the wave covector x - ray quanta components in the lagrange co - ordinates system having diffracted on this mirror are determined by formula @xmath389 .\ ] ] according to the definition we have for parameters @xmath390 and @xmath391 @xmath392 } ; \nonumber \\ & & \beta ( \xi ^{1},\xi ^{2})=-\frac{n_{3}^{({\bf l})}(\xi ^{1},\xi ^{2 } ) } { n_{3}^{({\bf 0})}(\xi ^{1},\xi ^{2 } ) } ; \nonumber \\ & & c(\xi ^{1},\xi ^{2})=n_{i}^{({\bf 0})}(\xi ^{1},\xi ^{2})g^{ij}(\xi ^{1},\xi ^{2})n_{j}^{({\bf l})}(\xi ^{1},\xi ^{2 } ) , \label{36}\end{aligned}\ ] ] where @xmath393 and @xmath394 are the wave covectors guiding cosines of x - ray quanta falling and diffracted on the distorted bragg mirror surface in the lagrange co - ordinates system : @xmath395 ; \nonumber \\ & & n_{i}^{({\bf l})}(\xi ^{1},\xi ^{2})=\frac{\left [ \sqrt{1+\alpha } n_{i}^ { ( { \bf l})}+n_{\mu } ^{({\bf 0})}\left ( \frac{\partial u^{\mu } ( { \bf \xi } ) } { \partial \xi ^{i}}\right ) _ { \xi ^{3}=0}+\left ( \frac{\partial \psi _ { 0}({\bf \xi } , \varepsilon ) } { \partial \xi ^{i}}\right ) _ { \xi ^{3}=0}\right ] } { l(\xi ^{1},\xi ^{2 } ) } , \nonumber \\ & & l(\xi ^{1},\xi ^{2})=\sqrt{\left [ \sqrt{1+\alpha } n_{i}^{({\bf l})}+n_{\mu } ^{({\bf 0})}\left ( \frac{\partial u^{\mu } ( { \bf \xi } ) } { \partial \xi ^{i } } \right ) _ { \xi ^{3}=0}+\left ( \frac{\partial \psi _ { 0}({\bf \xi } , \varepsilon ) } { \partial \xi ^{i}}\right ) _ { \xi ^{3}=0}\right ] g^{ij}(\xi ^{1},\xi ^{2 } ) } \nonumber \\ & & \times\sqrt{\left [ \sqrt{1+\alpha } n_{j}^{({\bf l})}+n_{\mu } ^{({\bf 0 } ) } \left ( \frac{\partial u^{\mu } ( { \bf \xi } ) } { \partial \xi ^{j}}\right ) _ { \xi ^{3}=0}+\left ( \frac{\partial \psi _ { 0}({\bf \xi } , \varepsilon ) } { \partial \xi
^{j}}\right ) _ { \xi ^{3}=0}\right ] } \nonumber\end{aligned}\ ] ] consequently the bragg reflection coefficient of the distorted mirror in this approximation looks like @xmath396 here @xmath397 .
the asymmetry parameter @xmath107 and the polarization factor defined by formulas ( [ 36 ] ) can be considered the same as for the plane mirror in the first approximation .
if besides we neglect the debye - waller dependency on the co - ordinates , the formula ( 37 ) for the bragg coefficient is extremely simplified : @xmath398 the fact , that the distorted bragg mirror with rather big average radius of curvature can be represented as a set of plane bragg mirrors turned relatively to each other , is widely used in x - ray optics systems designing .
nevertheless we havent found the formulas ( [ 37]-[38 ] ) introducing this representation ( or their analogues ) in other papers .
the linear dimensions of the area participating in the forming of the diffracted field in the point @xmath399 on the crystal surface are about the x - ray quanta crystal absorption length @xmath400 .
if the deformation field in this area can be considered as the constant value , the formulas ( [ 37]-[38 ] ) are quite applicable .
once the bragg reflection coefficient is obtained , we can define the field of x - ray quanta diffracted on the distorted bragg mirror reflecting surface @xmath401 \label{39}\end{aligned}\ ] ] the equations ( [ 37]-[38 ] ) and ( [ 39 ] ) solve the problem of x - ray quanta two - wave dynamic diffraction on the elastically deformed crystal in the bragg geometry , i.e. on the incurved bragg mirror .
the only problem remaining is the x - ray quanta propagation and focusing in vacuum ( or in the air ) , i.e. the hemholts equation solution .
@xmath402 with the assumption that on the distorted bragg mirror surface there is the x - ray quanta field @xmath403 defined with the amplitude @xmath404 and phase @xmath405 . beyond the caustics
this task can be solved by geometrical optics methods [ 17 ] leading to the formula @xmath406 where @xmath407 is the distance from the bragg mirror surface point with parameters @xmath408 to the point with co - ordinates @xmath409 , and @xmath410 and @xmath411 are the curvature radiuses of the wave diffracted on the bragg mirror surface defined by the equations @xmath412 here @xmath413 is the fully asymmetrical tensor of the third rank ( @xmath414 ) , @xmath415 is the normal vector to the x - ray quanta wave front on the bragg mirror surface in the euler co - ordinates system defined by the formula @xmath416 } { l(\xi ^{1},\xi ^{2 } ) } ; \nonumber \\ & & \left ( l(\xi ^{1},\xi ^{2})\right ) ^{2}=\left ( \hat{j}^{-1}(\xi ^{1},\xi ^{2})\right ) _ { j}^{i}\left [ \sqrt{1+\alpha } n_{i}^{({\bf l})}+n_{\mu } ^ { ( { \bf 0})}\left ( \frac{\partial u^{\mu } ( { \bf \xi } ) } { \partial \xi ^{i } } \right ) _ { \xi ^{3}=0}+\left ( \frac{\partial \psi _ { 0}({\bf \xi } , \varepsilon ) } { \partial \xi ^{i}}\right ) _ { \xi ^{3}=0}\right ] \nonumber \\ & & \times \delta ^{jh}\left ( \hat{j}^{-1}(\xi ^{1},\xi ^{2})\right ) _ { h}^{l } \left [ \sqrt{1+\alpha } n_{l}^{({\bf l})}+n_{\mu } ^{({\bf 0})}\left ( \frac { \partial u^{\mu } ( { \bf \xi } ) } { \partial \xi ^{l}}\right ) _ { \xi ^{3}=0}+\left ( \frac{\partial \psi _ { 0}({\bf \xi } , \varepsilon ) } { \partial \xi ^{l}}\right ) _ { \xi ^{3}=0}\right ] \label{42}\end{aligned}\ ] ] where @xmath417 is the matrix inverse to the jacob matrix @xmath418 and @xmath419 are the euler co - ordinates of the deformed crystal points . in the points where the equations @xmath420 are fulfilled , the geometric - optical solution ( 40 ) turns into the infinity ( has a singularity ) .
these points are called the focal ones and the set of the focal points is called the caustic .
consequently the multitude called the caustic is determined by the equations @xmath421 as the example illustrating formulas ( [ 37])-([43 ] ) we will consider a simple situation when the plane waves packet of x - ray quanta with parallel vectors and energies in the interval from @xmath422 to @xmath423 incident on the plane - parallel silicon plate with thickness @xmath350 and reflection vector @xmath424 perpendicular to the plate entrance surface .
we will consider the crystal plate oriented in such the way that the bragg condition precisely holds for the x - ray quanta with energies @xmath366 .
next this plate is distorted by momentums @xmath340 and @xmath341 uniformly distributed over the plate sides in such a way that the deformations field is determined by the formulas ( [ 34 ] ) from the previous part and the distorted bragg mirror is the result of this deformation shown on fig.1 . since by accepted condition
the packet of plane waves is incident on this mirror , one would let @xmath425 in formulas ( [ 37])-([43 ] ) in this situation .
the angle between the diffraction plate and the co - ordinate plate @xmath426 will be denoted as @xmath427 .
fig.7 - 10 shows the graphs obtained with the help of formula ( [ 38 ] ) for the module of coefficient of bragg reflection from this distorted mirror for x - ray quanta energies @xmath428 @xmath429 @xmath366 and @xmath423 with @xmath430 .
we remind that the caustics ( the focal point multitude ) of x - ray quanta diffracted on the bragg mirror is defined exclusively by the x - ray radiation field and does not depend on the reflected quanta intensity distribution ( bragg radiation coefficient ) on the crystal surface . by fig.11 - 14
we give the caustics for x - ray quanta energies @xmath422 , @xmath431 , @xmath366 and @xmath423 obtained from formulas ( [ 42]-[43 ] ) for the whole bragg mirror ( without accounting the intensity distribution of quanta reflected on the whole surface of the mirror , i.e. without accounting the bragg reflection coefficient dependency on the co - ordinates of the point on the mirror surface ) .
it is clear that in experiment there will be presented only the part of this caustic surface which x - ray quanta are coming at , once reflected from the mirror area with the bragg reflection coefficient sufficiently differing from null .
these areas for our example are shown on fig.5 .
fig.15 - 18 shows caustics ( focal points manifold ) obtained also with the help of formulas ( [ 41]-[43 ] ) but taking into account the dependency of bragg reflection coefficients on the co - ordinates .
it is worth mentioning that even in the simplest case we deal with rather complicated ( that is clear from these diagrams ) combination of elementary differentiable peculiarities ( catastrophes ) @xmath432 and @xmath433 ( by v.i .
arnold classification ) of phase function @xmath434 .
that is why one can not even talk about the fit ( frequently met in the x - ray optics articles , see for ex .
@xcite ) of x - ray quanta intensity distribution near the caustics ( focus ) with gaussian function @xcite .
the caustics intensity distribution is described by special catastrophes functions the simplest of them being airy functions and piersey integral @xcite investigated in details in articles @xcite . if one models the intensity distribution by gaussian beams , it is necessary to use the special method of gaussian beams summing discussed in @xcite .
we are indebted to prof .
s. maksimenko and prof .
g. slepyan for stimulating discussions .
kryukovskiy , d.s .
lukin , e.a .
palkin , uniform asymptotics of quickly oscillating functions with singular saddle points , a.sc .
ussr , radiotechnics and electronics institute , preprint # 41 ( 413 ) , m. , 1984
|
in this work the interaction of electromagnetic field with quasi - periodic media has been scrutinized .
we have obtained the formula for a distorted medium polarizability tenzor in the x - ray frequency band .
also there have been obtained the x - rays dynamic diffraction equations for the mediums with arbitrary smooth distortion . for these equations
the takagi - tuipin equations are obtained as a particular case .
we have got a very simple formula for the coefficient of x - rays reflection on the bent bragg mirror .
we have scrutinized the example of calculating the x - rays reflection and focusing by the bent bragg mirror .
also in this work one could find the rezults of calculating the reflected x - rays caustics form .
it is shown that there exists a possibility of linear dimensions of the x - rays focusing area having typical values of @xmath0 .
# 1@xmath1#1
|
spatially entangled biphotons , such as those generated by spontaneous parametric downconversion ( spdc ) , exhibit strong einstein - podolsky - rosen ( epr ) type correlations @xcite in the transverse position and transverse momentum degrees of freedom @xcite .
because these variables are continuous , the entanglement can be very high - dimensional with a typical schmidt number greatly exceeding @xmath4 @xcite .
this provides high information density which can be leveraged to increase channel capacity and security for quantum key distribution @xcite and dense coding @xcite .
other applications include ghost imaging @xcite , quantum computing @xcite , and quantum teleportation @xcite .
experimentally characterizing the spdc state is unfortunately difficult due to weak sources and low resolution detectors .
spatial entanglement is traditionally imaged by jointly raster - scanning photon - counting avalanche photodiodes ( apds ) to measure spatial correlations .
this scales extremely poorly with increasing detector resolution . with a biphoton flux of @xmath5 coincident detections per second
, it would take @xmath6 days to jointly scan a @xmath7 pixel region for a signal - to - noise ratio ( snr ) of @xmath8 . for @xmath9 pixels
, it would take @xmath10 days ( see eq .
[ eq : raster ] ) .
other approaches have been tried with mixed success .
intensified ccd cameras can measure the schmidt number @xcite , but do not detect single - photon correlations rendering them ineffective for most quantum applications .
arrays of photon counting apds could replace ccds , but they are currently low resolution , noisy and resource intensive ; especially since each pixel pair must be individually correlated @xcite . a recent , promising result averages intensity correlations over many images from a single - photon sensitive electron - multiplying ccd , reporting @xmath11 modes @xcite .
this technique is limited to a @xmath12 ms exposure time ( apds are sub - ns ) and is noisier than using apds because it does not isolate individual coincident detections . in ref @xcite , dixon _
al . _ reduce the number of measurements required for a raster - scan by only measuring in an area of interest where correlations are expected , reporting a channel capacity of @xmath13 bits per photon . while not a true full - field measurement , they highlight a critical feature of the spdc field . in _
both _ position and momentum representations , the distribution of correlations between pairs of detector pixels is very sparse despite dense ( not sparse ) single - particle distributions . applying ideas from the field of compressive sensing ,
we exploit prior knowledge of this sparsity to beat the `` curse of dimensionality '' @xcite and efficiently characterize the full biphoton field without raster scanning . in this article
, we implement a compressive sensing , photon - counting double - pixel camera that efficiently images single - photon spdc correlations in the near- and far - field at resolutions up to @xmath14 dimensions per detector . at @xmath9 resolution , the measurement time is reduced from @xmath10 days for raster scanning to around @xmath15 hours .
we perform an entropic characterization showing channel capacities of up to @xmath3 bits per photon , equivalent to @xmath16 independent , identically distributed modes .
sums of channel capacities in conjugate bases violate a epr steering bound @xcite by up to @xmath17 bits .
compressive sensing is a technique that employs optimization to measure a sparsely represented @xmath18-dimensional signal from @xmath19 incoherent measurements @xcite .
the approach is so - named because the signal is effectively compressed during measurement .
though sparsity is assumed , it is not known prior to measurement which elements contain appreciable amplitude .
compressive sensing must determine both which elements are significant and find their values . to detect a sparsely represented @xmath18-dimensional signal vector @xmath20
, we measure a series of @xmath19 values @xmath21 by multiplying @xmath20 by an @xmath22 sensing matrix @xmath23 such that @xmath24 @xmath25 where @xmath26 is a noise vector . because @xmath19 , this system is undetermined ; a given @xmath21 does not specify a unique @xmath20 .
the correct @xmath20 is recovered by minimizing a regularized least squares objective function @xmath27 where for example @xmath28 is the @xmath29 norm of @xmath30 and @xmath31 is a scaling constant .
the function @xmath32 is a regularization promoting sparsity .
common @xmath32 include @xmath20 s @xmath33 norm , assuming the signal is sparse , and @xmath20 s total variation , assuming the signal s gradient is sparse @xcite .
@xmath23 must be incoherent with the basis of interest , with the surprising and non - intuitive result that a random , binary sensing matrix works well .
given sufficiently large @xmath34 , the recovered @xmath20 approaches the exact signal with high probability @xcite . for a @xmath35sparse signal , the required @xmath34 scales as @xmath36 .
incoherent , random sampling is particularly beneficial for low - light measurements as each measurement receives on average half the total photon flux @xmath37 , as apposed to @xmath38 for a raster scan .
compressive sensing is now beginning to be used for quantum applications such as state tomography @xcite .
shabani et .
al , for example , perform a tomography on a two qubit photonic gate for polarization entangled photons @xcite .
cs has also been used to with spatially correlated light for ghost imaging @xcite .
it is important to note that for ghost imaging , cs is not required to recover the full two - particle probability distribution as in entanglement characterization .
the quintessential compressed sensing example is the single - pixel camera @xcite .
an object is imaged onto a digital micromirror device ( dmd ) , a 2d binary array of individually - addressable mirrors that reflect light either to a single detector or a dump .
rows of the sensing matrix @xmath23 consist of random , binary patterns placed sequentially on the dmd . for an @xmath18-dimensional image , minimizing eq .
[ eq : objective ] recovers images using as few as @xmath39 measurements .
the single - pixel camera concept naturally adapts to imaging correlations by adding a second detector .
consider placing separate dmds in the near - field or far - field of the spdc signal and idler modes , where `` on '' pixels are redirected to photon counting modules .
the signal of interest is @xmath40 where @xmath41 represents the probability of a coincident detection between the @xmath42 pixel on the signal dmd and @xmath43 pixel on the idler dmd . the functions @xmath44 and @xmath45 are approximate position and momentum wavefunctions for the biphoton @xmath46 subscripts @xmath47 and @xmath48 refer to signal and idler photons respectively , @xmath49 and @xmath50 are the pump and correlation widths , and @xmath51 is a normalizing constant .
@xmath52 of eq .
[ eq : objective ] is simply a one - dimensional reshaping of @xmath53 or @xmath54 . like the single - pixel camera , a series of random patterns
are placed on the dmds to form rows of @xmath23 . for each pair of patterns ,
correlations between the signal and idler photons form the measurement vector @xmath21 .
minimization of eq .
[ eq : objective ] recovers @xmath41 .
while a fully random @xmath55 is preferred , the dmds only act on their respective signal or idler subspace , which prevents arbitrary @xmath55 .
rows of @xmath55 are therefore outer products of rows of single - particle sensing sensing matrices @xmath56 and @xmath57 @xmath58 where rows of @xmath56 represent random patterns placed on the signal dmd and rows of @xmath57 represent random patterns placed on the idler dmd . to make signal and idler photons distinguishable , @xmath56 and @xmath57 are not the same .
the validity of kronecker - type sensing matrices has been established and is of current interest in the cs community as attention shifts to higher dimensional signals @xcite .
the measurement vector @xmath21 is obtained by counting coincident detections for the series of dmd configurations given by @xmath23 .
a variety of reconstruction algorithms exist for eq .
[ eq : objective ] , with their computational complexity dominated by repeatedly calculating @xmath59 and @xmath60 @xcite .
this is especially unwieldy for correlation measurements as the size of @xmath23 is @xmath61 for @xmath62pixel dmds . using properties for kronecker products @xcite ,
these can be more efficiently computed by @xmath63 where @xmath64 and @xmath65 reshape a vector to a square matrix and vice - versa ; @xmath66 forms a vector from the diagonal elements of a square matrix ; and @xmath67 forms a square matrix placing the operand vector on its diagonal .
the compressive approach finds the joint probability distribution orders of magnitude faster than raster scanning through two key improvements .
the first is simply the reduction in the number of measurements . to jointly raster scan an @xmath1-pixel space
requires @xmath68 measurements . for a compressive measurement , sparsity is approximately @xmath1 with dimensionality @xmath68 ,
so only @xmath69 measurements are required . in practice
, we found excellent results when @xmath34 was only three percent of @xmath68 .
the second advantage of compressive measurements is that they more efficiently use available flux . for the raster scan ,
the total flux is distributed over at best @xmath1 pairs of pixels in the case of perfect correlations .
conversely , the average flux per incoherent compressive measurement is independent of @xmath1 , with each measurement receiving on average @xmath70 the total flux . to maintain constant snr ( photons / measurement ) with increasing @xmath1 , total measurement time therefore scales as @xmath71 for raster scanning .
given a photon flux of @xmath72 photons per second , the measurement time for a desired snr is @xmath73 where @xmath74 is the time per measurement . for incoherent , compressive measurements , acquisition time scales as @xmath75 .
the compressive improvement therefore scales as @xmath0 . for @xmath76 ,
this is of order @xmath77 .
this scaling factor somewhat optimistically assumes the reconstruction process yields an accurate result despite a noisy signal .
unfortunately , propagation of uncertainty through the reconstruction process remains a difficult problem , especially for non - ideal , real world systems @xcite .
there has been much recent theoretical work on the topic for gaussian @xcite and poissonian noise @xcite .
these results tend to require ideal sensing matrices or more complicated formulations to give provable performance bounds .
as such , their findings are difficult to directly , quantitatively apply to experiment . however , they do reveal pertinent features that indicate cs can perform extremely well in the presence of noise .
a well known characteristic of cs is a rapid phase change from poor to good quality reconstructions @xcite .
this phase change is often discussed as a function of increasing @xmath78 , with the boundary @xmath79 .
a similar phase transition occurs as a function of the noise level ; in our case , this is the number of photons per measurement .
for some cases , these two are even linked @xcite .
a practical compressive measurement simply requires large enough @xmath78 and photon flux @xmath72 to be in the space of good reconstructions .
fortunately , simply obtaining a recognizable reconstruction generally indicates the measurement conditions exceed this threshold . unlike a direct measurement , the information obtained by a series of @xmath80 compressive measurements
is contained in their deviation from the average value @xmath81 . in the presence of noise
, these deviations must exceed the noise level .
assuming poissonian shot noise , good reconstructions require @xmath82 , where @xmath83 is the standard deviation of the measurement vector and @xmath84 is a positive constant greater than one .
mm and @xmath85 mm form a 4@xmath86 imaging system with the crystal and dmds placed in the object and image planes respectively . for momentum correlations
, f1 is removed and the dmd is placed in the focal plane of @xmath87 mm .
photons striking dmd `` on '' pixels are directed to large area , single - photon counting modules .
photon arrivals are then correlated by a coincidence circuit . ] and
the phase change behavior versus photon number can be clearly seen .
the experiment used @xmath89 total ( 1250 detected ) photons per measurement to comfortably exceed the phase change .
the mse approaches a value @xmath90 corresponding to an snr of roughly @xmath91.,title="fig : " ] ; , @xmath92 pixel simulations with @xmath88 measurements and @xmath93 photons per measurement .
the red , constant line gives the true mutual information for the simulated object .
the black points give the @xmath92 far field experimental data for comparison.,title="fig : " ] ; ) .
the solid line represents the threshold that must be _ exceeded _ to witness epr steering .
error bars enclose two standard deviations from the expected uncertainty from simulations ( not performed for @xmath94 ) . because simulations systematically underestimate the mutual information , values above the bound are convincing . ]
the particular algorithm chosen to solve eq .
[ eq : objective ] also plays a role in the reconstructions accuracy .
these often have provable performance on ideal signals , but degrade when confronted with noisy or otherwise non - ideal conditions . in these circumstances
, they have various strengths , including speed , accuracy , and sensitivity to user selected parameters such as @xmath31 in eq .
[ eq : objective ] . for more information on common reconstruction algorithms ,
see refs @xcite . in practice , the best way to determine accuracy for a particular signal , sensing matrix , and reconstruction approach remains repeated simulations or experiments . for our system , we reduce a @xmath95 measurement from a @xmath10 day raster scan ( snr of 10 ) to an @xmath15 hour compressive acquisition , a thousand - fold improvement .
the experimental apparatus is given in fig .
[ fig : setup ] .
light from a @xmath96 mw , @xmath97 nm hecd laser was directed to a @xmath98 mm long bibo crystal oriented for type i , collinear spdc .
the generated daughter photons passed through a @xmath99 nm narrowband filter before separating into signal and idler modes at a @xmath100 beamsplitter . to measure position - position correlations , lenses @xmath101 mm and @xmath102 mm imaged the crystal onto signal and idler mode dmds . for momentum - momentum correlations ,
@xmath103 was removed and the dmds were placed in the focal plane of @xmath104 mm .
dmd `` on '' pixels reflected light to large area , single photon counting modules ( spcm ) connected to a correlating circuit . to measure @xmath41 , a series of @xmath34 random patterns
were placed on the dmds to form the sensing matrix @xmath23 . for each set of patterns ,
joint detections were counted for acquisition times @xmath105 for a total measurement time @xmath106 to make up the measurement vector @xmath80 .
the joint distribution @xmath41 was reconstructed using a gradient projection solver for eq .
[ eq : objective ] with @xmath107 regularization , commonly referred to as basis pursuit denoising @xcite .
we measured at dimensions of @xmath108 , @xmath109 , and @xmath110 corresponding to dmd resolutions of @xmath111 , @xmath7 , and @xmath112 pixels .
the associated measurement numbers @xmath34 were @xmath11 , @xmath113 , and @xmath114 so that @xmath34 is only about @xmath115 .
acquisition times were @xmath98 second for position measurements and @xmath116 seconds for momentum measurements to average @xmath4 coincident detections per dmd configuration in all cases .
additionally , we performed representative simulations at @xmath111 and @xmath7 resolutions .
a simulation for measuring position - position correlations at @xmath117 dmd resolution is given in fig .
[ fig : simrecon ] . the object in [ fig : simrecon](a )
is the correlation function of eq .
[ eq : p ] .
the simulation used @xmath88 measurements and a photon flux of @xmath93 photons / measurement multiplied by the ideal @xmath41 , conditions representative of the @xmath98 second experimental acquisitions .
note that this is the total signal strength before interacting with the sensing matrix ; the mean value of the measurement vector is @xmath118 detected photons .
values of the measurement vector were poissonian distributed to simulate the effect of shot noise .
[ fig : recon](b ) gives the reconstructed correlation function @xmath119 ) between signal and idler dmd pixels .
the sharply defined diagonal line shows the expected positive correlations between the two dmds .
dmd pixels are listed in column - major order . the mean squared error ( mse ) for
the reconstruction was @xmath90 .
the two - dimensional signal marginal distribution is inset , which provides an image of the signal beam
[ fig : simrecon](c ) and [ fig : simrecon](d ) sum the result along the anti - diagonal to show the correlation width @xmath50 .
qualitatively , the reconstruction closely resembles the original object , faltering only near the edges of the distribution where the signal falls beneath a noise floor .
the reconstruction recovers @xmath120 pixel with negligible error . to demonstrate the reconstruction accuracy ,
simulations were performed for increasing photon flux @xmath72 with dmd resolution @xmath111 and @xmath88 .
the mse versus @xmath72 is given in fig .
[ fig : mse ] .
reconstructions were normalized to the incident flux @xmath72 for comparison to the ideal signal .
the result shows the rapid phase change from poor to excellent reconstructions with a mse converging to @xmath90 beyond the phase change .
the mse can be used to roughly estimate the signal - to - noise ratio for a particular measurement of an average , non - zero element . assuming perfect pixel correlations and uniform marginal distributions , the energy in the signal is distributed over @xmath121 elements .
the signal - to - noise ratio is then @xmath122 . for @xmath92 pixels and mse @xmath123
, this yields an approximate snr of @xmath91 . for comparison , using eq .
[ eq : raster ] , a raster scan would require about four days to achieve a snr of only @xmath8 .
the simulated cs acquisition time was @xmath124 minutes for @xmath11 , @xmath98 second measurements .
sample experimental reconstructions for position - position and momentum - momentum correlations at @xmath125 pixel dmd resolution are given in fig.[fig : recon ] .
as in the simulations , the position - position result shows a well defined diagonal line indicating positive pixel correlations .
conversely , the momentum - momentum result shows an anti - diagonal line showing the expected anti - correlations . figs .
[ fig : recon](c ) and [ fig : recon](d ) sum the results along the anti - diagonal ( position - position ) and diagonal ( momentum - momentum ) to reveal an effective correlation width @xmath126 of a single pixel .
our detection scheme is therefore as accurate as possible at this resolution and our channel capacity remains detector limited .
once @xmath41 is recovered , the channel capacity is given by the classical mutual information shared between signal and idler dmd pixels ; @xmath127 where for example , @xmath128 is the signal particle s marginal probability distribution @xcite .
this entropic analysis is solely measurement based and does not require reconstructing a wavefunction or density matrix , a challenging task even for low - dimensional systems @xcite . to estimate the uncertainty in the mutual information from shot noise and the reconstruction process , we performed @xmath129 simulations at @xmath92 pixel resolution and @xmath130 simulations at @xmath131 pixel resolution .
simulations were not performed at @xmath94 pixel resolution due to available computer time .
in addition to the results from the raw reconstruction , thresholding was performed to provide noise reduction , where all values in the recovered @xmath41 below a percentage of the maximum value are forced to zero .
the simulated mutual information versus thresholding percentage is given in fig .
[ fig : thresh ] for the @xmath92 pixel simulations exemplified by fig .
[ fig : simrecon ] .
errorbars enclose one standard deviation from repeated simulations .
as the threshold increases from zero , the mutual information rises as a weak , uncorrelated noise floor is removed .
an optimal threshold is quickly reached , beyond which the threshold removes more signal than noise , reducing the mutual information .
note that the reconstructed mutual information is systematically lower than the actual mutual information in the ideal object .
this is due to remaining noise and difficulty in recovering parts of the signal towards the tail of the distribution .
the @xmath92 far field experimental result is included for comparison to the simulation .
the experiment closely matches the simulation both for no thresholding and beyond its optimal threshold , but is smaller in the intermediate region .
this is likely due to experimental errors not included in the simulation .
these include slight pixel misalignment between signal and idler dmds , optical aberrations , detector dark noise , stray light , power fluctuations in the laser , and temperature stability of the nonlinear crystal .
[ fig : thresh ] indicates that these experimental difficulties appear to increase the uncorrelated noise floor rather than significantly affect the correlated part of the reconstruction . although thresholding is a simple post - processing technique , it is applicable to how the entangled pixels might be used for communication .
if a pair of entangled pixels has a correlated amplitude near or below the background noise , it will be difficult to use that particular pair for communication .
a communication scheme would likely only consider pixel pairs above a certain threshold to be useful .
this is related to the technique in photonic quantum information of subtracting background noise from a measured signal . in cs , it is also common to perform post - processing or secondary optimization after maximizing sparsity , such as the debiasing routine used in ref .
@xcite . the experimental channel capacity versus dmd resolution for both position - position and momentum - momentum is given in fig .
[ fig : info ] for several levels of thresholding .
the optimal threshold is that which maximizes the mutual information . at @xmath132 and @xmath133 pixel resolutions ,
optimal thresholds of @xmath134 and @xmath135 were used for position - position and momentum - momentum distributions respectively . at @xmath2 pixel resolution ,
noise was more significant , so the optimal thresholds increased to @xmath135 and @xmath136 .
error bars on @xmath92 and @xmath131 pixels measurements represent the expected effect of shot noise and reconstruction uncertainty derived from simulation .
these have been conservatively set to include two standard deviations from the simulated result .
the joint probability distribution was also fit to the double - gaussian wavefunction ( eq . [ eq : wavefunction ] ) to find effective widths @xmath126 and @xmath137 . when @xmath138 , the mutual information between particles for eq .
[ eq : wavefunction ] is the logarithm of the federov ratio @xcite @xmath139 where the ratio is squared for two dimensions .
while this technically applies to the continuous wavefunction , and the true @xmath50 is smaller than a dmd pixel , eq .
[ eq : federov ] still applies to the discritized measurement so long as the effective @xmath140 .
fitting yielded the largest channel capacities with a maximum of @xmath3 bits for momentum - momentum correlations at @xmath2 pixel resolution , equivalent to @xmath16 independent , identically distributed , entangled modes . given that fitting more accurately characterizes the system and gives a larger mutual information , it is reasonable to question the usefulness the direct mutual information computation .
however , the two approaches suit different purposes .
fitting is useful if one is particularly interested in the state itself .
however , if one intends to use correlated pixels for some other purpose , such as communication , the direct calculation is more appropriate .
this is because the correlated pixels on the low intensity , tail of the distribution will be difficult to use in practice even if their amplitude can be inferred by fitting .
the solid curve of fig . [ fig : info ] gives the maximum possible mutual information between two , @xmath1-pixel detectors . assuming perfect diagonal or anti - diagonal correlations and uniform marginals , this is simply @xmath141 .
because we have gaussian marginals , we do not expect to reach this bound , even with @xmath142 pixel . by magnifying and using only the central part of the field , we could approach this upper limit .
despite not reconstructing a full density matrix , it is still possible to demonstrate non - classical behavior by comparing position - position and momentum - momentum correlation measurements directly .
this has traditionally involved fitting the measurements to eq .
[ eq : wavefunction ] and analyzing products or sums of conditional variances @xcite .
we recently presented a more inclusive , entropic steering inequality for witnessing continuous variable entanglement with discrete measurements @xcite , where the sum of the classical mutual information between position - position and momentum - momentum correlations is classically bounded . for our system
, all classically correlated measurements must satisfy @xmath143 where @xmath144 and @xmath145 are the respective widths of dmd pixels in the position and momentum basis .
note that @xmath146 is simply the bandwidth product for the dmd area , and is independent of @xmath1 if total area does not change .
the sum of the classical mutual information in conjugate bases for each detector resolution is given in fig .
[ fig : epr ] .
the solid blue line provides right hand side of eq .
[ eq : steering ] , which must be exceeded to witness epr steering .
error bars for the @xmath92 and @xmath131 cases are derived from simulation and include two standard deviations . in all cases ,
we show epr steering both with optimal thresholding and fitting to the double gaussian wavefunction ( eq . [ eq : wavefunction ] ) .
even at @xmath147 thresholding , there is a violation for @xmath125 dimensions .
recall that simulations ( fig .
[ fig : mse ] ) systematically under - represented the object mutual information relative to measurement uncertainty , so measurement error is highly unlikely to have over - estimated this sum . for the fitted @xmath148 dimensional
result , we violate the classical bound by @xmath17 bits .
in this article , we present a compressive sensing , double - pixel camera for characterizing the spcd biphoton state with photon - counting detectors .
this technique is very efficient , improving acquisition times over raster - scanning by @xmath0 for @xmath62pixel detectors .
we image spdc correlations at up to 1024 dimensions per detector and measure a detector - limited mutual information of up to 8.4 bits .
we also violate an entropic epr steering bound , indicating that these correlations are non - classical .
more broadly , our results suggest that compressive sensing can be extremely effective for analyzing correlations within large dimensional signals ( eg .
intensity - intensity correlations ) .
potential applications range from verifying security in spectral correlations for energy - time qkd @xcite to imaging through scattering media @xcite .
54ifxundefined [ 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/physrev.47.777 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.92.210403 [ * * , ( ) ] _ _ , @noop ph.d .
thesis , ( ) link:\doibase 10.1103/physrevlett.67.661 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.96.090501 [ * * , ( ) ] link:\doibase 10.1103/physreva.77.062323 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.69.2881 [ * * , ( ) ] link:\doibase 10.1103/physreva.61.042302 [ * * , ( ) ] link:\doibase 10.1103/physreva.52.r3429 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.87.123602 [ * * , ( ) ] link:\doibase 10.1103/physreva.83.052325 [ * * , ( ) ] link:\doibase 10.1103/physreva.76.033801 [ * * , ( ) ] link:\doibase 10.1103/physreva.80.022307 [ * * , ( ) ] http://dx.doi.org/10.1364/opn.22.4.000026 [ * * , ( ) ] http://llwww.ll.mit.edu/publications/journal/pdf/vol13_no2/13_2laserradars3d.pdf [ * * , ( ) ] in \doibase doi:10.1117/12.861600 [ _ _ ] , vol .
( , ) p. link:\doibase 10.1038/ncomms1988
[ * * ( ) , 10.1038/ncomms1988 ] link:\doibase 10.1103/physrevlett.108.143603 [ * * , ( ) ] @noop _ _ ( , ) http://arxiv.org/abs/1210.4234 [ `` , '' ] ( ) , , link:\doibase 10.1109/tit.2006.871582 [ * * , ( ) ] http://stacks.iop.org/0266-5611/23/i=3/a=008 [ * * , ( ) ] link:\doibase 10.1109/msp.2007.4286571 [ * * , ( ) ] link:\doibase 10.1109/msp.2007.914731 [ * * , ( ) ] http://users.ece.gatech.edu/~justin/l1magic/ [ `` , '' ] ( ) link:\doibase 10.1109/tit.2005.862083 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.105.150401 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.106.100401 [ * * , ( ) ] link:\doibase 10.1063/1.3238296 [ * * , ( ) ] link:\doibase 10.1103/physreva.84.061804 [ * * , ( ) ] in link:\doibase 10.1109/icip.2006.312577 [ _ _ ] ( ) pp .
link:\doibase 10.1109/msp.2007.914730 [ * * , ( ) ] in link:\doibase 10.1109/icassp.2010.5495900 [ _ _ ] ( ) pp .
link:\doibase 10.1109/tip.2011.2165289 [ * * , ( ) ] http://www.caam.rice.edu/~optimization/l1/tval3/ [ `` , '' ] ( ) @noop _ _ ( , ) link:\doibase 10.1117/1.3596602 [ * * , ( ) ] link:\doibase 10.1109/tit.2011.2165823 [ * * , ( ) ] link:\doibase 10.1109/tit.2012.2205894 [ * * , ( ) ] in link:\doibase 10.1109/ciss.2012.6310927 [ _ _ ] ( ) pp .
in link:\doibase 10.1109/isit.2009.5205258 [ _ _ ] ( ) pp . in link:\doibase 10.1109/ssp.2009.5278495 [ _ _ ] ( ) pp .
link:\doibase 10.1103/physrevlett.104.188701 [ * * , ( ) ] link:\doibase 10.1109/jstsp.2007.910281 [ * * , ( ) ] link:\doibase 10.1109/tsp.2011.2161292 [ * * , ( ) ] link:\doibase 10.1103/physreva.40.2404 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.103.160505 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.106.130402 [ * * , ( ) ] http://stacks.iop.org/0953-4075/42/i=17/a=175503 [ * * , ( ) ] link:\doibase 10.1103/physreva.40.913 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.84.2722 [ * * , ( ) ] @noop * * link:\doibase 10.1103/physrevlett.98.060503 [ * * , ( ) ] link:\doibase 10.1364/ol.36.000394 [ * * , ( ) ]
|
we implement a double - pixel , compressive sensing camera to efficiently characterize , at high resolution , the spatially entangled fields produced by spontaneous parametric downconversion .
this technique leverages sparsity in spatial correlations between entangled photons to improve acquisition times over raster - scanning by a scaling factor up to @xmath0 for @xmath1-dimensional images .
we image at resolutions up to @xmath2 dimensions per detector and demonstrate a channel capacity of @xmath3 bits per photon . by comparing the entangled photons classical mutual information in conjugate bases , we violate an entropic einstein - podolsky - rosen separability criterion for all measured resolutions . more broadly , our result indicates compressive sensing can be especially effective for higher - order measurements on correlated systems .
|
the short range energy transfer phenomena find applications in wide areas of physics and chemistry from mesoscopic to molecular systems , @xcite and is critical to our understanding of physical , chemical and biological processes at the molecular level .
@xcite understanding the physics of an exciton transfer from a donor to an acceptor is also crucial for designing better solar cells and photo - voltaic materials .
this has fueled great interest in the study of the effect of electromagnetic environment on the resonant energy transfer , which offers the possibility of controlling and enhancing the energy transfer in molecular systems .
@xcite therefore , obtaining an exact classical description of the resonant energy transfer under the influence of an electromagnetic environment is imperative to improve the energy transfer mechanism in molecular systems .
the interaction between a dipole emitter and its mirror image under the metallic surface , or equivalently , the scattered evanescent waves from the metal , has been studied in varieties of nanoscale systems .
@xcite furthermore , the effect of metal on the luminescent lifetime of a molecule has been known , @xcite and recently , its effect on the relaxation of the semiconducting quantum dot has been studied .
@xcite the presence of metallic environment in the vicinity of the donor - acceptor system introduces a fluctuating electromagnetic field which can modify the donor s emission and acceptor s absorption spectra , as well as the coupling strength between the donor and the acceptor molecules , which can modify the frster resonant energy transfer ( fret ) rate .
@xcite as a result , one can control the transfer rate by tuning the dielectric environment .
the derivation of the classical energy transfer rate using the dyadic green s function in the local limit of the metallic environment has been considered before @xcite . here
we reconsider the derivation based on photon green s function , connecting it with previous derivation based on molecular polarizability .
we also consider the non - local limit of the material permittivity which is an accurate limit at small separation of the donor - acceptor system from the metallic surfaces .
furthermore , we generalize our derivation to include energy transfer processes in multimer ( multi - donor and acceptor ) systems in the presence of an arbitrary electromagnetic environment .
we emphasize that the usage of the photons green function allows us to treat arbitrary photonics / dielectric environment as well as multimer systems efficiently , from both computational and theoretical perspectives .
two key factors in the fret rate are : the electronic coupling of the donor and acceptor dipoles and the overlap of the emission and absorption spectra of donor and acceptor molecules .
the fret is considered as an incoherent energy transfer mechanism in which greater overlap between the emission and absorption spectra due to incoherent broadening can enhance the transfer rate .
an incoherent evanescent wave in the vicinity of the metallic environment can participate in the fret by bridging the emission and absorption spectra due to the shared modes of evanescent waves from the metal .
the similar concept of shared phonon modes has been studied previously @xcite in the context of the enhancement of the fret rate and recent experiments have shown longer coherence time and enhancement of the exciton transfer in light harvesting proteins .
@xcite in the same spirit of the shared mode approach , the mirror image of the donor dipole can couple to the donor and acceptor at the same time through the evanescent waves , leading to enhanced transfer rate .
in addition , in the multimer systems , the shared modes will generate additional pathways for the energy transfer and their interference effect can suppress or enhance the resonant energy transfer rate .
from the classical electromagnetic ( em ) theory , the energy flux density of the em field from the donor to acceptor is given by the poynting vector , @xmath0 .
adopting the classical perspective by kuhn and silbey , @xcite the energy transfer from donor to acceptor can be understood as two coupled oscillating dipoles ( donor and acceptor ) .
the donor radiates an electric field that permeates the acceptor and the acceptor as an antenna receives the energy from this field . due to the existence of an electromagnetically active environment like metal , the electric field experienced by the donor and the acceptor
will be modified and can be determined by the boundary conditions , @xmath1 the geometry of the metallic environment . here
we employ the dyadic green s function to calculate the modified electric field of a donor molecule due to the presence of metallic environment , the effect of which is fully captured by the relative permittivity @xmath2 of the environment . the retarded photon green s function @xmath3 satisfies the following wave equation @xcite : @xmath4 d_{ik}(\vec{r}_d,\vec{r}_a , \omega ) \nonumber \\ & = \delta^{3}\left(\vec{r}_d-\vec{r}_a\right ) \delta_{jk}\,.\end{aligned}\ ] ] here @xmath5 is the speed of light in vacuum and @xmath6 and @xmath7 are positions of donor and acceptor , respectively .
we briefly pause here to mention that the dyadic green s function contains two important points of information .
first , it provides the electric field @xmath8 of the donor dipole @xmath9 , where @xmath10 is the strength or magnitude of the electric dipole and @xmath11 is the unit dipole , located at @xmath6 in the presence of arbitrary metallic environment , @xmath12 \cdot \vec{n}^{elec}_d\ , .\end{aligned}\ ] ] for molecules with finite volumes that can not be approximated by point dipoles , the dipole electric field is obtained by integrating the dyadic green s function over the volume of the molecule .
second , it provides the thermal correlation of the photon reservoir , which includes both evanescent and propagating waves , governed by the spectral density @xmath13 @xmath14 which is related to the dyadic green s function via fluctuation - dissipation theorem : @xcite @xmath15 [ 1 + n(\pm\omega)]\,,\end{aligned}\ ] ] where @xmath16 is the planck s function and the indices @xmath17 run over cartesian coordinate axes @xmath18 and @xmath19 . with this information ,
we are now in a position to express the rate of energy transfer in terms of green s function .
we start from the previously known expression of the energy transfer rate per unit volume in the time domain : @xcite @xmath20 where @xmath21 is the polarization of an acceptor in the time domain and @xmath22 is the electric field due to the donor molecule at the acceptor s location @xmath7 .
to compare with the fret rate , we transform @xmath23 into the frequency domain , @xmath24 using the acceptor molecule s polarization @xmath25 , where @xmath26 is the polarizability tensor of the acceptor , the energy transfer rate becomes : @xmath27 which in time - domain reduces to the previous result derived by chance @xmath28
@xcite . using eq .
[ dgf2em ] together with @xmath29 , where @xmath11 is the unit electric dipole moment of the acceptor molecule and @xmath30 is the total volume , and dividing the energy transfer rate by @xmath31 , we obtain the rate of resonant energy transfer : @xmath32 here we defined the absorption spectrum of the acceptor molecule @xmath33 $ ] and the donor emission spectrum @xmath34 , which according to fermi s golden rule , is @xmath35 , where @xmath36 is the franck - condon factor and @xmath37 is the distribution function of the donor state @xmath38 such that @xmath39 . in our model , the effect of electromagnetic environment is fully captured by the photons green s function , which can be obtained computationally by solving maxwell s equation .
[ rate ] shows that the transfer rate is determined by the overlap of the donor emission spectrum @xmath40 , acceptor absorption spectrum @xmath41 and the dyadic green s function @xmath42 .
these three terms represent two fluctuation sources phonon bath ( linewidth broadening of emission and absorption spectra ) and the evanescent component of the fluctuating electromagnetic waves . using the dyadic green s function ,
we extend the fret rate from dimer to multimer systems . assuming that the system consists of @xmath43 with @xmath44 donors and @xmath45 with @xmath46 acceptors , one can write the dyadic green s function for all possible interaction between the donors and acceptors using the same photon green s function described above , except with different labels of @xmath47 for a donor molecule location and @xmath48 for the location of an acceptor molecule .
the multimer fret ( mfret ) rate is given by @xmath49 the electric field experienced by the @xmath50-th acceptor due to the @xmath51-th donor consists of a direct field originating from the @xmath51-th donor and an indirect field mediated by other donors and acceptors , due to their induced dipole moments by the @xmath51-th donor field : @xmath52\nonumber\end{aligned}\ ] ] here we considered the first order indirect field only .
however , we stress that our formalism can be generalized to accommodate higher order indirect fields .
the induced polarization of the @xmath50-th acceptor @xmath53 is given by @xmath54 , where @xmath55 .
similar to our previous derivation of the transfer rate in a dimer system , we first transform eq .
[ qmc ] into the frequency domain and then use eq .
[ edmc ] , along with the acceptor s polarization , to obtain the rate of resonant energy transfer in multimer systems : @xmath56^\dag[\vec{n}^{\dag elec}_{a_m } \cdot \hat{d}_{{a_m } , { d_j } } ( \omega ) \cdot \vec{n}^{elec}_{d_j } ] % \
, \nonumber \\ % & + \frac{2\omega^2}{\hbar \varepsilon_0 c^2 } \mbox{re}\bigg\{\sum_{l'\neq j } ^{n_d } \sigma_{a_m , d_{l'}}(\omega)\ , \sigma_{d_l , d_j}(\omega ) \times \nonumber \\ % & [ \vec{n}^{\dag elec}_{d_l } \cdot \hat{d}^\dag_{a_m , d_l}(\omega ) \cdot \vec{n}^{elec}_{a_m } ] [ \vec{n}^{\dag elec}_{a_m } \cdot \hat{d}_{a_m , d_{l'}}(\omega ) \cdot \vec{n}^{elec}_{d_{l ' } } ] \times \nonumber \\ % & [ \vec{n}^{\dag elec}_{d_{l ' } } \cdot \hat{d}_{d_{l ' } , d_j}(\omega ) \cdot \vec{n}^{elec}_{d_j } ] + \sum_{m'\neq m } ^{n_a } \sigma_{a_m , a_{m'}}(\omega ) \sigma_{d_l , d_j}(\omega ) \times \nonumber \\ % & [ \vec{n}^{\dag elec}_{d_l } \cdot \hat{d}^\dag_{a_m , d_l}(\omega ) \cdot \vec{n}^{elec}_{a_m } ] [ \vec{n}^{\dag elec}_{a_m } \cdot \hat{d}_{a_m , a_{m ' } } ( \omega ) \cdot \vec{n}^{elec}_{a_{m ' } } ] \nonumber \\ % & [ \vec{n}^{\dag elec}_{a_{m ' } } \cdot \hat{d}_{a_{m ' } , d_j}(\omega ) \cdot \vec{n}^{elec}_{d_j } ] \bigg\ } + \mathcal{o}(\chi^3(\omega ) ) \bigg]\end{aligned}\ ] ] here we considered terms that are of the order @xmath57 only and ignored all other higher order terms .
[ mcfret ] is equivalent to the previously derived result @xcite , where neither dielectric environment was considered nor green s function formalism was employed .
we emphasize that our formalism readily allows us to include other higher order terms in the transfer rate .
the higher order terms associated with the indirect field ( the first order indirect field is presented in eq .
[ edmc ] ) can potentially induce stronger path interference in multimer system of certain geometrical arrangement .
furthermore , eq . [ mcfret ] includes more channels / paths for energy transfer from donors to acceptors .
the interplay among different paths , phonon and photon baths can lead to enhancement or reduction of the transfer rate in such systems .
we start by considering a dimer donor - acceptor system to understand the distance dependent enhancement due to coupling of the dimer with the evanescent wave from the metallic surface using bulk material permittivity in the local limit .
we assume that the separation distance between the donor and the acceptor @xmath58 is smaller than the vacuum wavelength @xmath59 , that is , @xmath60 , and neglect any retardation effects .
we consider silver metal with plasma frequency @xmath61 rad / s , and the electron scattering rate @xmath62 rad / s .
@xcite the emission and absorption spectra of donor and acceptor molecules in vacuum are assumed to be gaussian functions centered around @xmath63 , and @xmath64 , respectively , with a standard deviation of one - fifth of the donor s emission frequency .
@xcite without loss of generality , we assume that both dipoles point in the @xmath65-direction .
we then compare the energy transfer rate from donor to acceptor in the presence and absence of metal .
the analytical expression of green s functions for planar geometries are presented in the app .
[ app1 ] . from donor to acceptor in the presence of silver half - space ( dashed red ) and silver thin film ( solid blue ) to the rate of energy transfer in vacuum @xmath66 as a function of distance @xmath67 from the metallic surface .
the donor - acceptor separation distance @xmath68 and thin film thickness @xmath69 . ] in fig .
[ f1 ] we plot the ratio of donor - acceptor energy transfer rate in the presence of metallic half - space and thin film to the rate of energy transfer in vacuum . here
, we assume a donor and an acceptor are located at a distance @xmath70 above the silver surface .
we find that the energy transfer rate is modified significantly in the presence of metallic environment at small separation distances from the metal . for distances
@xmath67 shorter than the donor - acceptor separations , @xmath71 , the energy transfer rate is strongly enhanced in the presence of metal . in the opposite limit , i.e. , @xmath72 , the transfer rate mediated by metallic environment is suppressed and it saturates to the vacuum transfer rate . at a small separation distance from the metal , there exists a high density of evanescent modes which is responsible for enhancement of the transfer rate in the presence of metal . since these modes decay exponentially with distance
, they have less effect on the transfer rate as we move further away from the metal .
[ f1 ] also reveals an interesting path interference effect in the resonant energy transfer process . in the presence of metallic environment ,
excitons have two pathways@xmath73direct way from the donor to the acceptor and indirect way mediated by metal@xmath73to transfer from the donor to the acceptor .
the destructive interference between these pathways leads to suppression of energy transfer around @xmath74 and the constructive interference leads to enhancement of the transfer rate around @xmath75 .
next , we vary the separation distance @xmath76 between the donor - acceptor molecules keeping the distance @xmath77 from the metallic surface fixed . the result is plotted in fig .
the metallic half - space and thin film result deviates significantly for separation @xmath76 greater than film thickness @xmath78 .
as the separation gets larger , the evanescent modes on the opposite face of the thin film also contribute to the coupling of donor - acceptor system , leading to enhancement of the transfer rate . from a donor to an acceptor in the presence of silver half - space ( dashed red ) and thin film ( solid blue ) to the rate of energy transfer in vacuum @xmath66 as a function of donor - acceptor separation distance @xmath79 .
the distance from the metallic surface is held constant at @xmath80 and the thin film thickness @xmath81 . ] in fig .
[ f3 ] , we fix the position of the donor at @xmath82 above the silver surface and vary the position of the acceptor in both @xmath83- and @xmath65- directions .
we find that the energy transfer rate is significantly modified when the acceptor is close to the silver surface , as shown in fig .
[ f3 ] . near the metal surface ,
the presence of high density of evanescent modes significantly increase the transfer rate . from a donor to an acceptor in the presence of thin silver film of thickness @xmath81 to the rate of energy transfer in the vacuum @xmath66 as a function of acceptor position @xmath84 and @xmath85 , keeping the donor position fixed at @xmath86 and @xmath87 from the silver surface .
] next , we investigate the effect of the dielectric environment on the transfer rate of a multimer ( multi - chromophore ) system .
we consider a linear chain of donor and acceptor molecules that are placed horizontally above the silver surface .
we assume that the exciton transfers from the left most donor to the right most acceptor in the presence of intermediate donor molecules in a linear chain of the form @xmath88 , where @xmath89 and @xmath90 represent donor and acceptor molecules , respectively .
the ratio of multimer fret rate is plotted in fig .
we find that the point where destructive interference occurs , shifts from a larger to a smaller distance from the metal surface as we increase the number of intermediate donor molecules while keeping the length of the chain fixed .
for such an arrangement , the distance between the donor and the acceptor molecules drops . consequently , the point where destructive interference occurs also shifts toward a smaller distance from the metal , as seen in fig .
[ f4 ] . in a multimer system of
a linear chain of two ( solid blue ) three ( dashed blue ) and six ( dashed blue ) dipoles pointing in the z - direction in the presence of thin silver film of thickness @xmath91 to the rate of energy transfer in the vacuum @xmath92 as a function of distance @xmath67 from the metallic surface .
the exciton transfers from the left - most donor to the right - most acceptor which are separated by the distance @xmath93 . ] from donor to acceptor in the presence of silver thin film of thickness @xmath94 in the local limit ( dashed blue ) and non - local limit of permittivity ( solid blue ) to the rate of energy transfer in vacuum @xmath66 as a function of distance @xmath67 from the metallic surface .
the donor - acceptor separation distance @xmath68 . ] in the above discussions , we assumed a local limit of the dielectric environment by considering the bulk value of the metal permittivity .
however , it is important to emphasize that the bulk permittivity does not necessarily describe the surface effect very accurately , especially when the separation distance between the metal and donor - acceptor system becomes smaller than the mean free path of the electron in the metal .
since the energy transfer rate is modified only at a small separation from the metal , it is necessary to take an accurate model of the permittivity of the metal to obtain the better estimate of the effect of the metallic environment on the transfer rate . to this end
, we calculate the transfer rate using the non - local expression of the permittivity of a metallic thin film presented in the app . [ app2 ] .
the results are plotted in fig .
[ f5 ] .
we find that the non - local result does not differ significantly from the local result , unlike the case of relaxation rate in semiconducting quantum dots , where non - local result deviates significantly from the local result at small distances from the metal .
@xcite we attribute this difference in behavior to the fact that in case of energy transfer , green s function is evaluated at the spatial locations of a donor and an acceptor , which are always separated by a finite distance .
however , in the case of relaxation rate calculation , green s function is always evaluated at the location of a single molecule or a quantum dot .
the presence of finite length scale in the case of energy transfer prohibits contribution from large momentum wavevectors of the evanescent modes in the local limit and does not lead to divergence of the transfer rate even at small separations from the metal . as such ,
the non - local limit only offers small corrections and the result does not deviate significantly from the local approximation .
we have studied the effect of metallic environment on the resonant molecular energy transfer using the classical model based on the dyadic green s function in local and non - local limits of the material permittivity .
our results demonstrate that the evanescent waves from the metallic halfspace and thin film can enhance the fret rate between the donor and the acceptor since they are simultaneously coupled to the same evanescent mode from the metal .
furthermore , we have extended our model to multimer systems in arbitrary dielectric environments .
the numerical results indicate that since all donors and acceptors are under the influence of the evanescent wave , the path interference through all the incoherent channels can enhance or suppress the energy transfer rate with different arrangements of donors and acceptors and their distances from the metal surface .
our work suggests that by utilizing metallic environments and different arrangements of dipoles / molecules , the resonant energy transfer rate can be modulated . in the future
, we would like to implement computational method based on our theoretical framework to improve fster s energy transfer rate by tailoring the metallic geometry .
in addition , further studies to identify optimal geometrical arrangement of donors and acceptors to guide the energy flow in nanoscale systems will find applications in designing nano - materials for efficient harvesting of solar light .
furthermore , the current model can be used , with proper extension , to study heat transfer in nanoscale systems .
a.p . would like to thank maxim g. vavilov for fruitful discussions .
a.p . and m.r
were supported as part of the anser center , and energy frontier research center funded by the us department of energy , office of science , office of basic energy sciences , under award no .
de - sc0001059 .
a.p . and m.r .
also partially supported by the center for bio - inspired energy science , an energy frontier research center funded by the u.s .
department of energy , office of science , basic energy sciences under award de - sc0000989 .
x.c . would like to thank profs .
chunming niu and yonghong cheng for support and encouragement .
x.c . has been supported by the start - up funding of xian jiaotong university and acknowledges financial support from state key laboratory of electrical insulation and power equipment , school of electrical engineering , xian jiaotong university .
in this section , we present a few dyadic green s function that are analytically tractable .
first , we consider a special case of a homogeneous medium with relative permittivity @xmath95 and @xmath96 , for which the dyadic green s function @xmath97 is given by : @xmath98\ , , \end{aligned}\ ] ] where @xmath99 is a @xmath100 by @xmath100 diagonal matrix and @xmath101 .
below we consider other inhomogeneous geometries , like half - space and thin films , for which it is possible to compute the dyadic green s function analytically .
such geometries are also useful if we limit ourselves to the situation where the separation of the donor - acceptor system from the metal surface is smaller than the radius of curvature of the surface so that the surface can be assumed to be flat . for a source point at @xmath102 , a field point at @xmath103 and defining a two dimensional vector @xmath104 , the scattering green s function for a half space or a thin film geometry with the material permittivity @xmath105 in both local and non - local limits , and the surrounding permittivity @xmath106 is given by : @xmath107 \,,\\ & d_{zz}(\vec{r } , \vec{r } ' , \omega)=\frac{i c^2}{8\pi^2\omega^2 } \int_{0}^{2\pi } d\theta \int_{0}^{\infty}\frac{p^3}{\varepsilon_1 q_1}dp \nonumber \\ & \times e^{iq_1(z+z ' ) + i \vec{p}\cdot\vec{\rho } } \,\ , r_{p}(p ) \,,\end{aligned}\ ] ] where @xmath108 is the transverse and @xmath109 is the z - component of the wave vector , with @xmath110 .
all other components of the dyadic green s function can be computed from these components . in the above expression ,
@xmath111 and @xmath112 are the fresnel reflection coefficients of the medium @xmath113 with respect to the medium @xmath114 .
for the half space geometry with the material permittivity of the medium 2 in the local limit , these coefficients are given by the following expressions : @xmath115 where @xmath116 .
similarly , for a thin film geometry of thickness @xmath78 and the local permittivity @xmath105 surrounded by another medium of permittivity @xmath106 on both sides of the thin film , one can also derive the fresnel reflection coefficients analytically @xcite .
the results are : @xmath117
in the non - local limit , when the permittivity of the medium 2 depends on a position , @xmath118 , the fresnel reflection coefficients of the half space geometry take different forms and are given by the following expressions@xcite : where @xmath120 and @xmath121 are longitudinal and transverse components of the permittivity given by the following expressions : @xmath122 here @xmath123 is the electron collision frequency , @xmath124 is the plasma frequency , and @xmath125 is the fermi velocity .
blum , c. ; zijlstra , n. ; lagendijk , a. ; wubs , m. ; mosk , a. p. ; subramaniam , v. ; vos , w. l. nanophotonic control of the frster resonance energy transfer efficiency .
lett . _ * 2012 * , 109 , 203601 - 5 .
lunz , m. ; gerard , v. a. ; gunko , y. k. ; lesnyak , v. ; gaponik , n. ; susha , a. s. ; rogach , a. l. ; bradley , a. l. surface plasmon enhanced energy transfer between donor and acceptor cdte nanocrystal quantum dot monolayers .
_ nano lett . _ * 2011 * , 11 , 3341 - 5 .
joulain , k. ; mulet , j .-
p . ; marquier , f. ; carminati , r. ; greffet , j. -j .
surface electromagnetic waves thermally excited : radiative heat transfer , coherence properties and casimir forces revisited in the near field . _
rep . _ * 2005 * , 57 , 59 - 112 .
ordal , m. a. ; bell , r. j. ; alexander , r. w. ; long , l. l. ; querry , m. r. optical properties of fourteen metals in the infrared and far infrared - al , co , cu , au , fe , pb , mo , ni , pd , pt , ag , ti , v , and w. _ appl . opt . _ * 1985 * , 24 , 4493 - 4499 .
jang , s. ; cheng , y .- c .
resonance energy flow dynamics of coherently delocalized excitons in biological and macromolecular systems : recent theoretical advances and open issues . _ wiley interdiscip .
sci . _ * 2013 * , 3 , 84 - 104 .
kolkowitz , s. ; safira , a. ; high , a. a. ; devlin , r. c. ; choi , s. ; unterreithmeier , q. p. ; patterson , d. ; zibrov , a. s. ; manucharyan , v. e. ; park , h. ; lukin , m. d. probing johnson noise and ballistic transport in normal metals with a single - spin qubit .
_ science _ * 2015 * , 347 , 1129 - 1132 .
|
the high density of evanescent modes in the vicinity of a metal leads to enhancement of the near - field frster resonant energy transfer ( fret ) rate .
we present a classical approach to calculate the fret rate based on the dyadic green s function of an arbitrary dielectric environment , and consider non - local limit of material permittivity in case of metallic halfspace and thin film . in a dimer system , we find that the fret rate is enhanced due to shared evanescent photon modes bridging a donor and an acceptor .
furthermore , a general expression for the fret rate for multimer systems is derived .
the presence of a dielectric environment and the path interference effect enhance the transfer rate , depending on the combination of distance and geometry .
|
this paper treats the problem of screening a @xmath3-variate sample for strongly and multiply connected vertices in the partial correlation graph associated with the the partial correlation matrix of the sample .
this problem , called hub screening , is important in many applications ranging from network security to computational biology to finance to social networks . in the area of network security ,
a node that becomes a hub of high correlation with neighboring nodes might signal anomalous activity such as a coordinated flooding attack . in the area of computational biology
the set of hubs of a gene expression correlation graph can serve as potential targets for drug treatment to block a pathway or modulate host response . in the area of finance a hub
might indicate a vulnerable financial instrument or sector whose collapse might have major repercussions on the market . in the area of social networks
a hub of observed interactions between criminal suspects could be an influential ringleader .
the techniques and theory presented in this paper permit scalable and reliable screening for such hubs . unlike the correlations screening problem studied in this paper considers the more challenging problem of partial correlation screening for variables with a high degree of connectivity .
in particular we consider 1 ) extension to the more difficult problem of screening for partial correlations exceeding a specified magnitude ; 2 ) extension to screening variables whose vertex degree in the associated partial correlation graph exceeds a specified degree .
the hub screening theory presented in this paper can be applied to structure discovery in @xmath3-dimensional gaussian graphical models , a topic of recent interest to statisticians , computer scientists and engineers working in areas such as gene expression analysis , information theoretic imaging and sensor networks @xcite@xcite,@xcite .
for example , the authors of @xcite propose a euclidean nearest neighbor graph method for testing independence in a gauss - markov random field model .
when specialized to the null hypothesis of spatially independent gaussian measurements , our results characterize the large @xmath3 phase transitions and specify a weak ( poisson - like ) limit law on the number of highly connected nodes in such nearest neighbor graphs for finite number @xmath4 of observations .
many different methods for inferring properties of correlation and partial correlation matrices have been recently proposed @xcite,@xcite , , , .
several of these methods have been contrasted and compared in bioinformatics applications @xcite , , @xcite similar to the ones we consider in sec .
[ sec : experiments ] .
the above papers address the _ covariance selection _
problem @xcite : to find the non - zero entries in the covariance or inverse covariance matrix or , equivalently , to find the edges in the associated correlation or partial correlation graph .
the problem treated and the solution proposed in this paper differ from those of these previous papers in several important ways : 1 ) as contrasted to @xcite our objective is to screen for vertices in the graph instead of to screen for edges ; 2 ) unlike our objective is to directly control false positives instead of maximizing a likelihood function or minimizing a matrix norm approximation error ; 3 ) our framework is specifically adapted to the case of a finite number of samples and a large number of variables @xmath6 ) ; 4 ) our asymptotic theory provides mathematical expressions for the p - value for each of the variables with respect to a sparse null hypothesis on the covariance ; 5 ) unlike lasso type methods like the hub screening implementation can be directly applied to very large numbers of variables without the need for initial variable reduction .
additional relevant literature on correlation based methods can be found in . for specified @xmath7 $ ] and @xmath8 ,
a hub is defined broadly as any variable that is correlated with at least @xmath1 other variables with magnitude correlation exceeding @xmath0 .
hub screening is accomplished by thresholding the sample correlation matrix or partial correlation matrix and searching for rows with more than @xmath1 non - zero entries .
we call the former _ correlation hub screening _ and the latter _ partial correlation hub screening_. the screening is performed in a computationally efficient manner by exploiting the equivalence between correlation graphs and ball graphs on the set of z - scores .
specifically , assume that @xmath4 samples of @xmath3 variables are available in the form of a data matrix where @xmath9 .
first the columns of the data matrix are converted to standard @xmath4-variate z - scores .
the set of @xmath3 z - scores uniquely determine the sample correlation matrix .
if partial correlations are of interest , these z - score are replaced by equivalent modified z - scores that characterize the sample partial correlation matrix , defined as the moore - penrose pseudo - inverse of the sample correlation matrix . then an approximate k - nearest neighbor ( ann ) algorithm is applied to the z - scores or the modified z - scores to construct a ball graph associated with the given threshold @xmath0 .
hub variables are discovered by scanning the graph for those whose vertex degree exceeds @xmath1 .
the ann approach only computes a small number of the sample correlations or partial correlations , circumventing the difficult ( or impossible ) task of computing all entries of the correlation matrix when there are millions ( or billions ) of variables .
state - of - the - art ann software has been demonstrated on over a billion variables @xcite and thus our proposed hub screening procedure has potential application to problems of massive scale .
we also note that using the standard moore - penrose inverse is well understood to be a sub - optimal estimator of the partial correlation matrix in terms of minimum mean square error @xcite . to our knowledge
its properties for screening for partial correlations has yet to be investigated .
this paper demonstrates that its potential use in detecting high partial correlations as compared to estimating them
. no screening procedure would be complete without error control .
we establish limiting expressions for mean hub discovery rates .
these expressions are used to obtain an approximate phase transition threshold @xmath2 below which the average number of hub discoveries abruptly increases . when the screening threshold @xmath0 is below @xmath2 the discoveries are likely to be dominated by false positives .
we then show that the number of discoveries is dominated by a random variable that converges to a poisson limit as @xmath0 approaches @xmath10 and @xmath3 goes to infinity .
thus the probability of making at least one hub discovery converges to a poisson cumulative distribution . in the case of independent
identically distributed ( i.i.d . )
elliptically distributed samples and sparse block diagonal covariance matrix , the mean of the dominating variable does not depend on the unknown correlations . in this case we can specify asymptotic p - values on hub discoveries of given degree under a sparse - covariance null model .
finite @xmath3 bounds on the poisson p - value approximation error are given that decrease at rates determined by @xmath3 , @xmath1 , @xmath0 , and the sparsity factor of the covariance matrix .
to illustrate the power of the hub screening methods we apply them to a public gene expression datases : the nki breast cancer data @xcite .
each of these datasets contains over twenty thousand variables ( genes ) but many fewer observations ( genechips ) .
the screening reveals interesting and previously unreported dependency structure among the variables . for the purposes of exploring neighborhood structure of the discoveries we introduce a waterfall plot of their approximate p - values that plots the family of degree - indexed p - value curves over the range of correlation thresholds
this graphic rendering can provide insight into the structure and significance of the correlation neighborhoods as we sweep the variables over different vertex degree curves in the waterfall plot .
the outline of this paper is as follows . in sec .
[ sec : parcorscreening ] we formally define the hub screening problem . in sec .
[ sec : zscorerep ] we present the z - score representation for the pseudo - inverse of the sample correlation matrix . in sec .
[ sec : theory ] we give an overview of the results pertaining to phase transition thresholds and establish general limit theorems for the familywise discovery rates and p - values .
section [ sec : statementasythms ] gives the formal statements of the results in the paper .
the proofs of these results are given in the appendix . in sec .
[ sec : experiments ] we validate the theoretical predictions by simulation and illustrate the application of hub screening to gene microarray data .
let the @xmath3-variate @xmath11^t$ ] have mean @xmath12 and non - singular @xmath13 covariance matrix @xmath14 .
we will often assume that @xmath15 has an elliptically contoured density : @xmath16 for some non - negative strictly decreasing function @xmath17 on @xmath18 .
the correlation matrix and the partial correlation matrix are defined as @xmath19 and @xmath20 , respectively , where for a square matrix @xmath21 , @xmath22 denotes the diagonal matrix obtained from @xmath21 by zeroing out all entries not on its diagonal .
available for observation is a @xmath23 data matrix @xmath24 whose rows are ( possibly dependent ) replicates of @xmath15 : @xmath25=[\bx_{(1)}^t , \cdots , \bx_{(n)}^t]^t,\ ] ] where @xmath26^t$ ] and @xmath27 $ ] denote the @xmath28-th column and row , respectively , of @xmath24 .
define the sample mean of the @xmath28-th column @xmath29 , the vector of sample means @xmath30 $ ] , the @xmath13 sample covariance matrix @xmath31 , and the @xmath32 sample correlation matrix @xmath33 for a full rank sample correlation matrix @xmath34 the sample partial correlation matrix is defined as @xmath35 in the case that @xmath34 is not full rank this definition must be modified .
several methods have been proposed for regularizing the inverse of a rank deficient covariance including shrinkage and pseudo - inverse approaches . in this paper
we adopt the pseudo - inverse approach and define the sample partial correlation matrix as @xmath36 where @xmath37 denotes the moore - penrose pseudo - inverse of @xmath34 .
let the non - negative definite symmetric matrix @xmath38 be generic notation for a correlation - type matrix like @xmath39 , @xmath40 , @xmath41 , or @xmath42 .
for a threshold @xmath7 $ ] define @xmath43 the undirected graph induced by thresholding @xmath44 as follows .
the graph @xmath45 has vertex set @xmath46 and edge set @xmath47 , where an edge @xmath48 exists in @xmath49 if @xmath50 .
the degree of the @xmath28-th vertex of @xmath43 is @xmath51 , the number of edges that connect to @xmath28 .
we call @xmath52 the population correlation or partial correlation graph depending on whether @xmath44 is defined as @xmath53 or @xmath54 .
likewise we call @xmath43 the empirical correlation or partial correlation graph depending on whether @xmath55 or @xmath56 .
a @xmath13 matrix is said to be row - sparse of degree @xmath57 , called the _ sparsity degree _ , if no row contains more than @xmath58 non - zero entries .
when @xmath44 is row - sparse of degree @xmath57 its graph @xmath43 has no vertex of degree greater than @xmath57 .
a special case is a block - sparse matrix of degree @xmath57 ; a matrix that can be reduced via row - column permutation to block diagonal form with a single @xmath59 block .
assume that a single treatment of the @xmath3 variables yields data matrix @xmath60 of dimension @xmath23 .
a given vertex @xmath28 is declared a hub screening discovery at degree level @xmath1 and threshold level @xmath0 if |\ { j : ji , |_ij| } | , [ eq : parcorscreen ] where @xmath44 is equal to @xmath34 for correlation hub screening or is equal to @xmath61 for partial correlation hub screening .
we denote by @xmath62 the total number of hub screening discoveries .
to be practically useful , we need guidelines for selecting the threshold screening parameters in ( [ eq : parcorscreen ] ) . in the sequel
we will develop a large @xmath3 asymptotic analysis to address the following issues : 1 ) phase transitions in the number of discoveries as a function of these screening parameters ; 2 ) relations between the false positive rate of the discoveries and these screening parameters .
define the @xmath63 matrix of z - scores associated with the data matrix @xmath60 = [ _ 1 , ,
_ p]=(n-1)^-1/2 ( _ n- n^-1 1 1^t ) x ^-1/2 _ , [ eq : ordap ] where @xmath64 is the @xmath65 identity matrix and @xmath66^t\in \reals^{n}$ ] .
this z - score matrix is to be distinguished from the @xmath67 z - score matrices @xmath68 and @xmath69 , denoted collectively by the notation @xmath70 in the sequel , that are derived from the matrix @xmath71 .
we exploit the following z - score representation of the sample correlation matrix = ^t , [ eq : orda ] and defined a set of equivalent but lower dimensional z - scores called u - scores .
the u - scores lie in the unit sphere @xmath72 in @xmath73 and are obtained by projecting away the rowspace components of @xmath71 in the direction of vector @xmath74 .
more specifically , they are constructed as follows .
define the orthogonal @xmath75 matrix @xmath76 $ ] .
the matrix @xmath77 can be obtained by gramm - schmidt orthogonalization and satisfies the properties @xmath78 , \;\ ;
\bhn^t\bhn= \bi_{n-1}.\ ] ] the u - score matrix @xmath79 $ ] is obtained from @xmath71 by the following relation = ^t .
[ eq : udef ] assume that @xmath80 .
the moore - penrose pseudo - inverse of @xmath34 has the representation ^ = ^t[^t]^-2 .
[ lemma : mp ] the proof of the lemma simply verifies that @xmath81^{-2 } \mu$ ] satisfies the moore - penrose conditions for @xmath82 to be the unique pseudo - inverse of @xmath34 : 1 ) the matrices @xmath83 and @xmath84 are symmetric ; 2 ) @xmath85 ; and 3 ) @xmath86 . define the @xmath87 matrix of partial correlation z - scores @xmath88 $ ] , @xmath89 : = [ ^t]^-1 _ ^t[^t]^-2 ^ -1/2 .
[ eq : parcorscore]with this definition lemma [ lemma : mp ] gives the following z - score representation for the sample partial correlation matrix = ^t .
[ eq : pscore ]
there is a phase transition in the number of discovered variables as a function of the applied screening threshold(s ) .
this phase transition threshold , which we call @xmath90 , is such that if the screening threshold @xmath0 decreases below @xmath90 , the mean number @xmath91 $ ] of hub discoveries of degree @xmath1 abruptly increases to @xmath3 .
an expression for the critical threshold is obtained from the asymptotic expression ( [ eq : endef ] ) for the mean given in prop .
[ prop : parcor1 ] in the appendix _
c,= , [ eq : rhocrit ] where @xmath92and @xmath93 with @xmath94 the beta function .
the unknown covariance matrix @xmath14 influences @xmath90 only through the quantity @xmath95 , defined in ( [ eq : jdef ] ) , which is a measure of average @xmath96-order dependency among the z - scores @xmath97 . for large @xmath3
the constant @xmath98 depends only weakly on @xmath3 and the critical threshold increases to @xmath10 at rate @xmath99 , which is close to logarithmic in @xmath3 for large @xmath4 but much faster than logarithmic for small @xmath4 . when the rows of @xmath60 are i.i.d .
elliptically distributed and @xmath14 is block - sparse of degree @xmath57 then , from prop .
[ prop : prep ] j()=1+o((k / p)^ _ ) , [ eq : jpdasy ] where @xmath100 for correlation hub screening and @xmath101 for partial correlation hub screening .
when @xmath3 is large @xmath102 goes to zero and the phase transition threshold @xmath90 becomes independent of @xmath14 . as @xmath103 prop .
[ prop : parcor1 ] in the appendix states that @xmath104 is dominated by an asymptotically poisson random variable @xmath105 if : 1 ) @xmath106 increases to one at a prescribed rate depending on @xmath4 ; 2 ) the sparsity degree increases only as @xmath107 ; and 3 ) the dependency coefficient @xmath108 , defined in ( [ eq : deltapij ] ) , converges to zero .
this guarantees that @xmath109 converges to the poisson probability @xmath110 where @xmath111 is the rate parameter of @xmath105 .
the rate of convergence is provided in prop .
[ prop : parcor ] along with a finite @xmath3 approximation for @xmath111 @xmath112 with _ , = _
p pp-1 p_0(_p , n)^ , [ eq : lambdadef ] and @xmath113 is the spherical cap probability defined in ( [ eq : podef ] ) .
the parameter @xmath114 does not depend on the underlying distribution of the observations .
furthermore , when the rows of @xmath60 are i.i.d . elliptically distributed with covariance that is block - sparse of degree @xmath57 , prop .
[ prop : prep ] asserts that ( [ eq : jpdasy ] ) holds and that the dependency coefficient @xmath108 is equal to zero for correlation hub screening and of order @xmath115 for partial correlation hub screening .
therefore , when the block - sparse model is posed as the null hypothesis , prop .
[ prop : parcor1 ] implies that the false positive fwer error rate can be approximated by p(n_,>0)1-(- _ , ) .
[ eq : fpr ] approximate p - values can also be obtained under the block - sparse null hypothesis .
assume that for an initial threshold @xmath116 $ ] the sample correlation or partial correlation graph @xmath117 has been computed for @xmath44 equal to @xmath34 or @xmath61 .
consider a vertex @xmath28 in this graph that has degree @xmath118 .
for each value @xmath119 of the degree threshold @xmath1 denote by @xmath120 the maximum value of the correlation threshold @xmath0 for which this vertex maintains degree at least @xmath1 in @xmath121 .
@xmath122 and is equal to the sample correlation between @xmath123 and its @xmath1-th nearest neighbor .
we define the approximate p - value associated with discovery @xmath28 at degree level @xmath1 as pv_(i)1-(-_,_i ( ) ) .
[ eq : ppvalue ] the quantity ( [ eq : ppvalue ] ) is an approximation to the probability of observing at least one vertex with degree greater than or equal to @xmath1 in @xmath43 under the nominal block - sparse covariance model .
the accuracy of the approximations of false positive rate ( [ eq : fpr ] ) and p - value ( [ eq : ppvalue ] ) are specified by the bound ( [ eq : pvlim ] ) given in prop .
[ prop : parcor ] .
corollary [ cor : parcor ] provides rates of convergence under the assumptions that @xmath124 and the rows of @xmath60 are i.i.d . with sparse covariance .
for example , assume that the covariance is block - sparse of degree @xmath57 .
if @xmath57 does not grow with @xmath3 then the rate of convergence of @xmath109 to its poisson limit is no worse than @xmath125 for @xmath126 . on the other hand ,
if @xmath57 grows with rate at least @xmath127 , for @xmath128 , the rate of convergence is no worse than @xmath129 .
this latter bound can be replaced by @xmath130 for correlation hub screening under the less restrictive assumption that the covariance is row - sparse .
more generally the combination of prop .
[ prop : parcor ] and the assertions ( prop .
[ prop : prep ] ) that @xmath131 and @xmath132 yields cc o ( \{(k / p)^ _ , p^-(-1)/ ( k / p)^-1,p^-1/,(1-)^1/2 } ) , & > 1 + o ( \{(k / p)^ _ , ( k / p)^2 , p^-1,(1-)^1/2 } ) , & = 1 . [ eq : pvlim_main ] .
the terms @xmath133 , @xmath134 and @xmath135 respectively quantify the contribution of errors due to : 1 ) insufficient sparsity in the covariance or , equivalently , the correlation graph ; 2 ) excessive dependency among neighbor variables in this graph ; 3 ) poor convergence of @xmath91 $ ] ; and 4 ) inaccurate mean - value approximation of the integral representation of @xmath136 $ ] by ( [ eq : thetamoms ] ) .
one of these terms will dominate depending on the regime of operation .
for example , specializing to partial correlation hub screening ( @xmath100 ) , if @xmath137 and @xmath138 then @xmath139 and the deficiency in the poisson probability approximation will not be the determining factor on convergence rate .
as defined above , the approximate p - values provide a useful statistical summary .
rank ordering and thresholding the list @xmath140 of p - values at any level @xmath141 $ ] yields the set of vertices of degree at least @xmath1 that would pass a test of significance at false positive fwer level @xmath142 .
additional useful information can be gleaned by graphically rendering the aggregate lists of p - values as described below .
specifically , assume that correlation screening generates an associated family of graphs @xmath143}$ ] .
let @xmath144 be the maximum discovered degree in the initial graph @xmath145 .
we define the _ waterfall plot of p - values _ as the family of curves , plotted against the thresholds @xmath120 , indexed by @xmath146 where the @xmath1-th curve is formed from the ( linearly interpolated ) ranked ordered list of p - values @xmath147 , @xmath148 ( see fig .
[ fig : waterfallplot ] ) .
a useful visualization of the evolution of vertex neighborhoods over the family @xmath149}$ ] is the _ p - value trajectory _ over the waterfall plot .
this trajectory is defined as the ordered list @xmath150 defined for a given vertex @xmath28 .
all p - value trajectories start at the outermost curve ( curve associated with @xmath151 ) on the waterfall plot and extinguish at some inner curve ( associated with increasing @xmath137 ) .
vertices with the tightest large neighborhoods will tend to have long trajectories that start in the middle of the outer curve and extinguish at a curve deep in the waterfall plot , e.g. , the variable labeled arrb2 in fig .
[ fig : waterfallplot ] , while vertices with the tightest small neighborhoods will tend to have short trajectories that start near the extremal point of the outer curve , e.g. , the variable labeled ctag2 in fig .
[ fig : waterfallplot ] whose trajectory extinguishes near the bottom right corner of waterfall plot .
the asymptotic theory for hub discovery in correlation and partial graphs is presented in the form of three propositions and one corollary .
[ prop : parcor ] gives a general bound on the finite sample approximation error associated with the approximation of the mean and probability of discoveries given in prop .
[ prop : parcor1 ] .
the results of props . [ prop : parcor ] and [ prop : parcor1 ] apply to general random matrices of the form @xmath152 where the @xmath3 columns of @xmath70 lie on the unit sphere @xmath153 and , in view of ( [ eq : udef ] ) and ( [ eq : pscore ] ) , they provide a unified theory of hub screening for correlation graphs and partial correlation graphs .
corollary [ cor : parcor ] specializes the bounds presented in prop .
[ prop : parcor ] to the case of sparse correlation graphs using prop .
[ prop : prep ] . for @xmath154 , @xmath7 $ ] , and @xmath44 equal to the sample correlation matrix @xmath34 or the sample partial covariance matrix @xmath61 we recall the definition of @xmath104 as the number of vertices of degree at least @xmath1 in @xmath155 .
define @xmath156 as the count of the number of groups of @xmath1 mutually coincident edges in @xmath121 .
is equivalent to the number of subgraphs in @xmath157 that are isomorphic to a star graph with @xmath1 vertices .
] in the sequel we will use the key property that @xmath158 if and only if @xmath159 . for @xmath160 , @xmath7 $ ] , and @xmath161 define = _ p , n,,j ( ) [ eq : lambdadef ] where _ p , n,,=pp-1 p_0^ [ eq : xidef ] , @xmath162 is defined in ( [ eq : podef ] ) , @xmath163 is given in ( [ eq : jdef ] ) , and @xmath164 is the average joint density given in ( [ eq : olfdef2 ] ) .
we also define the following quantity needed for the bounds of prop . [
prop : parcor ] _
p,=p^1/(p-1 ) p_0 .
[ eq : etadef ] note that @xmath165 to order @xmath166 , where @xmath167 .
let @xmath168 be the function equal to @xmath10 for @xmath137 and equal to @xmath169 for @xmath151 .
let @xmath170 $ ] be a @xmath67 random matrix with @xmath171 .
fix integers @xmath1 and @xmath4 where @xmath154 and @xmath161 .
let the joint density of any subset of the @xmath172 s be bounded and differentiable . then
, with @xmath173 defined in ( [ eq : lambdadef ] ) , |e[n _ , ] - | o ( _ p,^ \{_p , p^-1/ , ( 1-)^1/2 } ) [ eq : enlim ] furthermore , let @xmath174 be a poisson distributed random variable with rate @xmath175=\lambda/\varphi(\delta)$ ] . if @xmath176 then , for any integer @xmath57 , @xmath177 , & & |p(n_,>0)-p(n^ * _ , >
0)|[eq : pvlim ] + & & \ { cc o(_p,^ \{_p,^ ( k / p)^+1 , q_p , k , , _
p , n , k,_1,p^-1/,(1-)^1/2 } ) , & > 1 + o(_p,1 \{_p,1 ( k / p)^2 , _ p , n , k,1_1 , p^-1,(1-)^1/2 } ) , & = 1 .
, with @xmath178 and @xmath108 defined in ( [ eq : deltapdefavg ] ) .
[ prop : parcor ] the proof of the above proposition is given in the appendix .
the poisson - type limit ( [ eq : poissonconv ] ) is established by showing that the count @xmath179 of the number of groups of @xmath1 mutually coincident edges in @xmath180 converges to a poisson random variable with rate @xmath181 .
let @xmath182 $ ] be a sequence converging to one as @xmath183 such that @xmath184 . then _
pe[n_,_p ] = _ n , _
p j ( ) , [ eq : endef ] where @xmath185 .
assume that @xmath107 and that the weak dependency condition @xmath186 is satisfied .
then p(n_,_p>0)1-(-/ ( ) ) .
[ eq : poissonconv ] [ prop : parcor1 ] the proof of prop .
[ prop : parcor1 ] is an immediate and obvious consequence of prop .
[ prop : parcor ] and is omitted .
propositions [ prop : parcor ] and [ prop : parcor1 ] are general results that apply to both correlation hub and partial correlation hub screening under a wide range of conditions .
corollary [ cor : parcor ] specializes these results to the case of sparse covariance and i.i.d .
rows of @xmath60 having elliptical distribution .
these are standard conditions assumed in the literature on graphical models .
in addition to the hypotheses of prop .
[ prop : parcor1 ] assume that @xmath187 and that the rows of @xmath24 are i.i.d .
elliptically distributed with a covariance matrix @xmath14 that is row - sparse of degree @xmath57 .
assume that @xmath57 grows as @xmath188 where @xmath189 .
then , for correlation hub screening the asymptotic approximation error in the limit ( [ eq : poissonconv ] ) is upper bounded by @xmath190 . under the additional assumption that the covariance is block - sparse , for partial correlation hub screening
this error is upper bounded by @xmath191 .
[ cor : parcor ] the proof of corollary [ cor : parcor ] is given in the appendix .
the proposition below specializes these results to sparse covariance .
let @xmath60 be a @xmath63 data matrix whose rows are i.i.d .
realizations of an elliptically distributed @xmath3-dimensional vector @xmath15 with mean parameter @xmath12 and covariance parameter @xmath14 .
let @xmath192 $ ] be the matrix of correlation z - scores ( [ eq : udef ] ) and @xmath193 $ ] be the matrix of partial correlation z - scores ( [ eq : parcorscore ] ) defined in sec .
[ sec : zscorerep ] .
assume that the covariance matrix @xmath14 is block - sparse of degree @xmath194 .
then the pseudo - inverse partial correlation matrix @xmath195 has the representation = ^t(1+o(q / p ) ) .
[ prep]let @xmath172 denote @xmath196 or @xmath197 and assume that for @xmath154 the joint density of any distinct set of z - scores @xmath198 is bounded and differentiable over @xmath199 .
then the @xmath96-fold average function @xmath163 ( [ eq : olfdef2 ] ) and the dependency coefficient @xmath200 ( [ eq : deltapdefavg ] ) satisfy j()=1+o((q / p)^ _ ) , [ eq : prep2 ] _ p , n , k,_1 = \ { cc o((q / p ) ) , & = 1 + 0 , & = 0 .
[ eq : prep3 ] where @xmath201 and @xmath202 for correlation and partial correlation hub screening , respectively , and @xmath203 .
[ prop : prep ] [ prop : parcorblocksparse ] _ proof of proposition [ prop : prep ] : _ the proof of proposition [ prop : prep ] is given in the appendix .
figure [ fig : pv1 ] shows the waterfall plot of partial correlation hub p - values for a sham measurement matrix with i.i.d .
normal entries that emulates the nki experimental data studied in the next subsection .
there are @xmath204 samples and @xmath205 variables in this sham . using ( [ eq : rhocrit ] ) with parameter @xmath206 , the critical phase transition threshold on discoveries with positive vertex degree was determined to be @xmath207 . for purposes of illustration of the fidelity of our theoretical predictions we used an initial screening threshold equal to @xmath208 .
as this is a sham , all discoveries are false positives . to illustrate the fidelity of the theoretical predictions waterfall plots of @xmath3-values ( [ eq : ppvalue ] ) are shown in fig .
[ fig : pv1 ] . for clarity , the figure shows the _
@xmath209-value _ defined as @xmath210 . when presented in this manner the leftmost point of each curve in the waterfall plot occurs at approximately @xmath211)$ ] , as can be seen by comparing the second and third columns of table 1 .
this table demonstrates strong agreement between the predicted ( mean ) number of partial correlation hub discoveries and the actual number of discoveries for a single realization of the data matrix .
the realization shown in the table is representative of the many simulations performed by us .
has @xmath204 rows and @xmath205 columns and is populated with i.i.d .
zero mean and unit variance gaussian variables ( @xmath212 ) .
dots on the curves correspond to unnormalized log p - values ( called @xmath209-values ) of discovered variables whose partial correlations exceeed the initial screening threshold @xmath208 .
each curve is indexed over a particular vertex degree threshold parameter @xmath1 ranging from @xmath151 to @xmath213 , the maximum vertex degree found .
legends indicate the total number of variables discovered having vertex degree @xmath214 .
left : plot of @xmath215 .
right : same as right panel but @xmath209-values are plotted on log scale .
, title="fig:",height=188 ] has @xmath204 rows and @xmath205 columns and is populated with i.i.d .
zero mean and unit variance gaussian variables ( @xmath212 ) .
dots on the curves correspond to unnormalized log p - values ( called @xmath209-values ) of discovered variables whose partial correlations exceeed the initial screening threshold @xmath208 . each curve
is indexed over a particular vertex degree threshold parameter @xmath1 ranging from @xmath151 to @xmath213 , the maximum vertex degree found .
legends indicate the total number of variables discovered having vertex degree @xmath214 . left : plot of @xmath215 .
right : same as right panel but @xmath209-values are plotted on log scale . , title="fig:",height=188 ] [ table : sham ] .fidelity of the predicted ( mean ) number of false positives and the observed number of false positives in the realization of the sham nki dataset experiment shown in fig .
[ fig : pv1 ] [ cols="^,^,^",options="header " , ] . the netherlands cancer institute ( nki ) dataset @xcite contains data from affymetrix genechips collected from 295 subjects who were diagnosed with early stage breast cancer .
in peng _ et al _ a graphical lasso method for estimating the partial correlation graph was proposed and was applied to this dataset . here
we apply partial correlation hub screening . .
the genes plotted correspond to vertices of positive degree in the initial partial correlation graph with threshold @xmath216 .
each curve indexes the p - values for a particular degree threshold @xmath1 and a gene is on the curve if its degree @xmath217 in the initial graph is greater than or equal to @xmath1 .
the discovered vertex degree ranges from @xmath10 to @xmath218 ( last dot labeled il14 at bottom left ) .
the p - value trajectories across vertex degree @xmath1 are indicated for several genes of interest .
note that all three affymetrix array replicates of igl@xmath219 were discovered . , height=377 ] but showing the p - value trajectories of @xmath220 of the hub genes ( bub1 cenpa knsl6 stk12 rad54l id - gap ) reported in peng _
these are the genes reported in table 4 of that have unambiguous annotation on the genechip array .
, height=377 ] as in peng @xmath221 we only used a subset of the available genechip samples . specifically , since 29 of the 295 genechips had variables with missing values , only 266 of the them were used in our analysis .
each genechip sample in the nki dataset contains the expression levels of 24,481 genes .
peng @xmath221 applied univariate cox regression to reduce the number of variables to 1,217 genes prior to applying their sparse partial correlation estimation ( space ) method .
in contrast , we applied our partial correlation hub screening procedure directly to all 24,481 variables .
an initial threshold @xmath222 was selected .
figure [ fig : waterfallplot ] illustrates the waterfall plot of p - values of all discovered variables .
note in particular the very high level of significance of certain variables at the lower extremities of the p - value curves .
according to ncbi entrez several of the most statistically significant discovered genes on these strands have been related to breast cancer , lymphoma , and immune response , e.g. arrb2 ( _ arrestin , beta 2 _ ) , ctag2 ( _ cancer / testis antigen _ ) , il14 ( _ interleukin _ ) , and igl@xmath219 ( _ immunoglobin alpha _ ) . the p - value trajectories ( colored labels ) across different values of @xmath1 of these four genes is illustrated in the figure .
note that some genes are statistically significant only at low vertex degree ( ctag2 ) while others retain high statistical significance across all vertex degrees ( igl@xmath219 ) .
fig [ fig : waterfallplot2 ] is the same plot with the trajectories of the @xmath220 unambiguously annotated hub genes given in table 4 of peng . while these @xmath220 genes do not have nearly as high p - values , or as high partial correlation , as compared to other genes shown in fig .
[ fig : waterfallplot ] their p - values are still very small ; less than @xmath223 .
we treated the problem of screening for variables that are strongly correlated hubs in a correlation or partial correlation graph when @xmath5 and @xmath3 is large .
the proposed hub screening procedure thresholds the sample correlation or the pseudo - inverse of the sample correlation matrix using z - score representations of the correlation and partial correlation matrices . for large @xmath3 and finite @xmath4
asymptotic limits that specify the probability of false hub discoveries were established .
these limits were used to obtain expressions for phase transition thresholds and p - values under the assumption of a block - sparse covariance matrix . to illustrate the wide applicability of our hub screening results we applied it to the nki breast cancer gene expression dataset . * acknowledgements * alfred hero was partially supported by national science foundation grant ccf 0830490 and digiteo .
bala rajaratnam was supported in part by nsf grants dms-05 - 05303 , dms-09 - 06392 and grant sufsc08-sushstf09-smscvisg0906 . 10 a. anandkumar , l. tong , and a. swami . .
, 55(2):816827 , 2009 .
r. arratia , l. goldstein , and l. gordon . .
, 5(4):403424 , 1990 .
bickel and e. levina .
covariance regularization via thresholding .
, 34(6):25772604 , 2008 .
chang , d.s.a .
nuyten , j.b .
sneddon , t. hastie , r. tibshirani , t. srlie , h. dai , y.d .
vant veer , h. bartelink , et al . .
, 102(10):3738 , 2005 .
cox and n. wermuth . .
chapman & hall / crc , 1996 .
dempster . .
, 28(1):157175 , 1972 .
j. friedman , t. hastie , and r. tibshirani . . , 2007
j. friedman , t. hastie , and r. tibshirani . .
, 9(3):432 , 2008 .
r. gill , s. datta , and s. datta . .
, 11(1):95 , 2010 .
m. goldstein and a. f. m. smith .
ridge - type estimators for regression analysis . , 36(2):pp .
284291 , 1974 .
g. h. golub and c. f. van loan . .
the johns hopkins university press , baltimore , 1989 .
a.o . hero and b. rajaratnam .
large scale correlation screening . , 2010 .
h. jgou , m. douze , and c. schmid . . , 2010
nicole krmer , juliane schfer , and anne - laure boulesteix .
regularized estimation of large - scale gene association networks using graphical gaussian models .
, 10(384):124 , 2009 .
y. liu , v. chandrasekaran , a. anandkumar , and a.s .
willsky . .
in _ information theory proceedings ( isit ) , 2010 ieee international symposium on _ , pages 16831687 .
ieee , 2010 .
j. peng , p. wang , n. zhou , and j. zhu . .
, 104(486):735746 , 2009 .
v. pihur , s. datta , and s. datta . . ,
24(4):561 , 2008 .
b. rajaratnam , h. massam , and c.m .
carvalho . .
, 36:28182849 , 2008 .
a. j. rothman , p.j .
bickel , e. levina , and j. zhu .
sparse permutation invariant covariance estimation .
, 2:494515 , 2008 .
j. schaefer and k strimmer . a shrinkage approach to large - scale covariance matrix estimation and implications for functional genomics . , 4(32 ) , 2005 .
a. wiesel , y.c .
eldar , and a.o . hero . .
, 58(3):14821492 , 2010 .
this appendix contains two subsections .
section [ sec : notation ] gives the necessary definitions .
section [ sec : asythms ] gives proofs of the theory given in sec .
[ sec : statementasythms ] . * @xmath60 : @xmath23 matrix of observations . *
@xmath224 $ ] : @xmath67 matrix of correlation ( or partial correlation ) @xmath225-scores @xmath226 associated with @xmath60 . * @xmath152 : @xmath13 sample correlation matrix @xmath34 ( if @xmath227 ) or sample partial correlation matrix @xmath61 ( if @xmath228 associated with @xmath60 .
* @xmath229 : sample correlation ( or partial correlation ) coefficient , the @xmath230-th element of @xmath152 . *
@xmath231 $ ] : screening threshold applied to matrix @xmath152 . *
@xmath232 : spherical cap radius parameter .
* @xmath72 : unit sphere in @xmath73 .
* @xmath233 : surface area of @xmath72 . *
@xmath52 : graph associated with population correlation matrix @xmath234 or partial correlation matrix @xmath235 .
an edge in @xmath52 corresponds to a non - zero entry of @xmath44 .
* @xmath236 : graph associated with thresholded sample correlation matrix @xmath237 or partial correlation matrix @xmath56 .
specifically , the edges of @xmath43 are specified by the non - diagonal entries of @xmath152 whose magnitudes exceed level @xmath0 . *
@xmath217 : observed degree of vertex @xmath28 in @xmath43 , @xmath238 . * @xmath1 : screening threshold for vertex degrees in @xmath43 , @xmath238 .
* @xmath57 : upper bound on vertex degrees of @xmath52 , @xmath239 .
* @xmath104 : generic notation for the number of correlation hub discoveries ( @xmath240 ) or partial correlation hub discoveries ( @xmath241 ) of degree @xmath242 in @xmath243 , or @xmath244 , respectively . *
@xmath156 counts the number of subsets of @xmath1 mutually coincident edges in @xmath157 .
* @xmath245 : the union of two anti - polar spherical cap regions in @xmath246 of radii @xmath247 centered at points @xmath248 and @xmath249 . * @xmath250 : probability that a uniformly distributed vector @xmath251 falls in @xmath245 + p_0 = p_0(,n)= & _ n _ ^1(1-u^2)^ du + = & ( n-2)^-1 _ n ( 1-^2)^(n-2)/2(1+o(1-^2 ) ) , + [ eq : podef ] where @xmath93 and @xmath252 is the beta function . for given integer @xmath57 , @xmath253 , and @xmath254 either the population correlation matrix @xmath53 or the population partial correlation matrix @xmath255 define _ k(i)=_j_1 j_(k,_i ) _
l=1^(k,_i ) |_ij_l| [ eq : nki ] , where @xmath256 denotes the degree of vertex @xmath28 in @xmath257 and the maximization is over the range of distinct @xmath258 that are not equal to @xmath28 . when @xmath259 these are the indices of the @xmath256 neighbors of vertex @xmath28 in @xmath257 .
when @xmath260 these are the subset of the @xmath57-nearest neighbors ( @xmath57-nn ) of vertex @xmath28 .
for the sequel it will be convenient to define the following vector valued indexing variable : @xmath261 , where @xmath262 and @xmath263 are distinct integers in @xmath264 . with this index denote by @xmath265 the set of @xmath266 z - scores @xmath267 .
define the set of complementary @xmath57-nn s of @xmath265 as @xmath268 , where a_k ( ) = ( _ l=0^_k(i_l))^c -\ { } [ eq : akij ] , with @xmath269 denoting set complement of set @xmath270 .
the complementary @xmath57-nn s include vertices outside of the @xmath57-nearest - neighbor regions of the set of points @xmath265 .
define the @xmath1-fold leave - one - out average of the density , a function of @xmath28 , @xmath271 : & & ( _ 1 , , _ , _ i ) [ eq : avgfubivpersistent ] [ eq : olfdef1 ] + & & = 2 ^ -d _ s_1 , , s_\{-1,1 } p-1 ^-1 _ i_1 i_i^p f__i_1 , , _ i_,_i(s_1 _ 1 ,
, s _ _ , _ i ) , where in the inner summation , indices @xmath272 range over @xmath264 . also define the @xmath96-fold average of the same density & & ( _ 1 ,
, _ , _ i ) [ eq : olfdef2 ] + & & = p^-1_i=1^p ( ( _ 1 , , _ , _ i ) + ( _ 1 , , _ , -_i ) ) . for any data matrix @xmath70 define the dependency coefficient between the columns @xmath265 and their complementary @xmath57-nn s _ p , n , k,()= ( f__|_a_k()-f__)/f _ _ _ , [ eq : deltapij ] and the average of these coefficients is _ p , n , k,_1= ( pp-1 ) ^-1_i_0=1^p _ i_1 < < i _ _ p , n , k , ( ) .
[ eq : deltapdefavg ] where the second sum is indexed over @xmath273 .
the coefficient ( [ eq : deltapdefavg ] ) quantifies weak dependence of the z - scores . if , for all @xmath28 , @xmath265 and its complementary @xmath57-nn neighborhood variables are independent then @xmath274 . when the rows of @xmath60 are i.i.d . and elliptically distributed , and @xmath227 are the standard correlation @xmath225-scores , then a sufficient condition for @xmath274 is that @xmath52 have no vertex of degree greater than @xmath57 or , equivalently , that the dispersion matrix @xmath14 be row sparse of degree @xmath57 .
with @xmath277 we have @xmath278 .
define @xmath279 the indicator of the presence of an edge in @xmath43 between vertices @xmath28 and @xmath280 , where @xmath281 is the union of two antipolar caps in @xmath72 of radius @xmath232 centered at @xmath282 and @xmath283 , respectively .
then @xmath284 and @xmath285 have the explicit relation _ i&=&_l=^p-1 _
_ i(p-1,l ) _ j=1^l _ ik_j _ m = l+1^p-1 ( 1-_i k_m ) [ eq : phirep ] where we have defined the index vector @xmath286 and the set @xmath287 the inner summation in ( [ eq : phirep ] ) simply sums over the set of distinct indices not equal to @xmath28 that index all @xmath288 different types of products @xmath289 . subtracting @xmath290 from both sides of ( [ eq : phirep ] ) & & _ i- _ _ i(p-1 , ) _ j=1^_ik_j + & & = _ l=+1^p-1 _ _ i(p-1,l ) _ j=1^l _ ik_j_m = l+1^p-1 ( 1-_i k_m ) + & & + _ _ i(p-1,l ) _
m = l+1^p-1(-1)^m - l _
< k_m _ j=1^l _ ik_j _ n = l+1^m _ ik_n [ eq : phirep2 ] where , in the last line we have used the expansion @xmath291
the following simple asymptotic representation will be useful in the sequel .
for any @xmath292 , @xmath293 , @xmath294 , e&=&_s_n-2 d_a(r , ) d_1 _ a(r , ) d_k f__i_1 , , _
i_k,_i(_1 , , _ k , ) [ eq : thetamoms ] + & & p_0^k
a_n^k m_k|1 [ eq : thetamomsineq ] [ eq : old714 ] with @xmath162 defined in ( [ eq : podef ] ) , @xmath233 , and m_k|1 & = & _ i_1 i_k+1 f__i_1 , , _ i_k|_i_k+1 _ [ eq : mk1def ] , the following simple generalization of ( [ eq : old714 ] ) to arbitrary product indices @xmath285 will also be needed ep_0^m
m_|q| , [ eq : thetamoms2 ] where @xmath295unique@xmath296 is the set of unique indices among the distinct pairs @xmath297 and @xmath298 is a bound on the joint density of @xmath299 . define the random variable _
i = p-1 ^-1 _ _ i(p-1 , ) _ j=1^_ik_j .
[ eq : thetadefnew ] we show below that for sufficiently large @xmath3 |e[_i]-p-1 e[_i]| & & _ p , ( ( p-1 ) p_0)^+1 , [ eq : proof1 ] where @xmath300 and @xmath301 is a least upper bound on any @xmath302-dimensional joint density of the variables @xmath303 conditioned on @xmath172 . to show inequality ( [ eq : proof1 ] ) take expectations of ( [ eq : phirep2 ] ) and apply the bound ( [ eq : old714 ] ) to obtain & & |e[_i]-p-1 e[_i]| + & & |_l=+1^p-1 p-1 l p_0^l a^lm_l|1 + p-1 _ l=1^p-1- p-1-l p_0^+l a_n^+l m_+l|1 | + & & a(1+(!)^-1 ) , [ eq : proof2 ] where @xmath304 the line ( [ eq : proof2 ] ) follows from the identity @xmath305 and a change of index in the second summation on the previous line . since @xmath306 |a| & & _ + 1l < p \{a_n^l m_l|1 } _
l=+1^p-1 p-1 l ( ( p-1)p_0)^l + & & _ + 1l
< p \{a_n^l m_l|1}(e-_l=0^ ) ( ( p-1)p_0)^+1 . application of the mean - value - theorem to the integral representation ( [ eq : thetamoms ] ) yields |e[_i]- p_0^j ( ) | & & _ p , ( ( p-1)p_0)^r , [ eq : proof3 ] where @xmath307 and @xmath308 is a bound on the norm of the gradient @xmath309 combining ( [ eq : proof1])-([eq : proof3 ] ) and the relation @xmath310 , |e[_i]- p-1 p_0^ j()| & & o(((p-1)p_0)^ \ { ( p-1)p_0 , ( 1-)^1/2 } ) .
[ eq : proof3pp ] summing over @xmath28 and recalling the definitions ( [ eq : xidef ] ) and ( [ eq : etadef ] ) of @xmath311 and @xmath312 , |e[n_,]-_p , n , , j()| & & o(p((p-1)p_0)^ \ { ( p-1)p_0 , ( 1-)^1/2 } ) + & = & o(_p,^ \{_p , p^-1/ , ( 1-)^1/2 } ) .
[ eq : proof3p ] this establishes the bound ( [ eq : enlim ] ) . for the bound ( [ eq : pvlim ] ) we use the chen - stein method @xcite .
the part of the bound ( [ eq : pvlim ] ) that holds for @xmath151 was derived in the course of proof of prop . 1 in .
below we treat the case @xmath137 .
recalling the definition @xmath156 as the number of subsets of @xmath1 mutually coincident edges in @xmath157 , we have the representation : _
, = _ i_0=1^p _
i_1 < < i__j=1^_i_0i_j [ eq : rep1 ] , where the second sum is indexed over @xmath313 . for @xmath314
define the index set @xmath315 where @xmath316 @xmath317 .
these index the distinct sets of points @xmath318 and their respective @xmath57-nn s . note that @xmath319 . identifying @xmath320 and
@xmath105 a poisson distributed random variable with rate @xmath321 $ ] , the chen - stein bound ( * ? ? ?
1 ) is 2_a |p ( _ , a)-p(n_,^*a)| b_1+b_2+b_3 , [ eq : chenstein ] where @xmath322e\left[\prod_{m=1}^{\delta}\phi_{j_0j_m}\right],\ ] ] @xmath323,\ ] ] and , for @xmath324 $ ] , @xmath325\right].\ ] ] over the range of indices in the sum of @xmath326 @xmath327 $ ] is of order @xmath328 , by ( [ eq : thetamoms2 ] ) , and therefore @xmath329 which follows from definition ( [ eq : etadef ] ) .
more care is needed to bound @xmath330 due to the symmetry relation @xmath331 .
if in the summations defining @xmath330 , @xmath332 and @xmath333 occur for some @xmath334 then there will be a match and @xmath335 .
in such case the summand of @xmath330 will be of lower order than @xmath336 .
for example , for the case that @xmath337 a match implies @xmath338 and , from ( [ eq : thetamoms2 ] ) , @xmath339=e[\prod_{l=1}^{\delta}\phi_{j_1 i_l}\prod_{m=2}^{\delta}\phi_{i_1 j_m } ] = o\left(p_0^{2\delta-1}\right).\ ] ] over @xmath340 and @xmath341 there can be no more than a single match in @xmath330 s summand . for a given match
there are at most @xmath342 summands of reduced order .
we conclude that b_2 & & o(p^+1 k^+1 p_0 ^ 2 ) + o(p^+1 k^-1 p_0 ^ 2 - 1 ) + & = & o(_p,^2(k / p)^+1)+ o(_p,^2 - 1(k / p)^-1 p^-(-1)/ ) , which follows from the relation @xmath343 .
next we bound the term @xmath344 in ( [ eq : chenstein ] ) .
the set @xmath345 indexes the complementary @xmath57-nn s of @xmath265 so that , using the representation ( [ eq : thetamoms2 ] ) , b_3 & = & _ ^ < e + & = & _ ^ < _ s_n-2^|a_k()| d_a_k ( ) ( _ l=1^ _ s_n-2
d_i_0 _ a(r,_i_0 ) d_i_l ) ( ) f__(_)f__a_k()(_a_k ( ) ) + & & o(p^+1 p_0^ _ p , n , k,_1 ) = o(_p,^_p , n , k,_1).observe that , with @xmath346 $ ] |p(n _ , > 0)-(1-(-))| & & |p ( _ , > 0)-p(n _ , > 0)| + & & + |p ( _ , > 0)-(1-(-e[_,]))| + & & + |(-e[_,])-(-)| + & & b_1+b_2+b_3 + o(|e[_,]-| ) .
[ eq : allterms ] combining the above inequalities on @xmath326 , @xmath330 and @xmath344 yields the first three terms in the argument of the `` max '' on the right side of ( [ eq : pvlim ] ) .
it remains to bound the term @xmath347-\lambda|$ ] .
application of the mean value theorem to the multiple integral ( [ eq : thetamoms2 ] ) gives |e- p_0^ j(f__i_1 , , _ i_,_i)| & & o(p_0^ r ) .
[ eq : phi_ijdbnd ] applying relation ( [ eq : rep1 ] ) |e [ _ , ] - pp-1 p_0^ j()| & & o(p^+1 p_0^ r ) = o(_p,^ r ) .
[ eq : phi_ijdbnd2 ] combine this with ( [ eq : allterms ] ) to obtain the bound ( [ eq : pvlim ] ) .
this completes the proof of prop .
[ prop : parcor ] . for correlation hub screening ( @xmath227 )
@xmath274 so it suffices to consider the other arguments of `` max '' in the bound ( [ eq : pvlim ] ) .
as in the proof of prop [ prop : parcor1 ] , @xmath348 and we can merge the last two terms in ( [ eq : pvlim ] ) into the single term @xmath349 , with @xmath142 defined in the corollary statement . finally , note that @xmath350 and @xmath351 when @xmath352 .
therefore , as @xmath353 when @xmath187 , we conclude that if @xmath354 all arguments of `` max '' in the bound ( [ eq : pvlim ] ) are dominated by @xmath355.turning to partial correlation hub screening ( @xmath356 ) , under the block - sparse covariance assumption prop .
[ prop : prep ] asserts that @xmath357 which dominates @xmath355 .
this completes the proof of cor .
[ cor : parcor ] . by block - sparsity
, the matrix @xmath68 of z - scores can be partitioned as @xmath358 $ ] , where @xmath359 $ ] and @xmath360 $ ] are the dependent and independent columns of @xmath68 , respectively . since the columns of @xmath361 s are i.i.d . and
uniform over the unit sphere @xmath72 , as @xmath103 we have @xmath362= ( n-1)^{-1 } \bi_{n-1}.\ ] ] furthermore , as the entries of the matrix @xmath363 are bounded by 1 , @xmath364 where @xmath365 is an @xmath366 matrix whose entries are of order @xmath367 .
hence , as @xmath368 , the pseudo - inverse of @xmath34 has the asymptotic large @xmath3 representation ^ =( ) ^2 ^t[_n-1+(q / p)]^-2 =( ) ^2 ^t ( 1+o(q / p ) ) , [ rdaggerrep ] which establishes ( [ prep ] ) .
define the partition @xmath369 of the index set @xmath370 where @xmath371 is the set of @xmath372-tuples restricted to the dependent columns @xmath373 of @xmath68 .
the summation representations ( [ eq : olfdef2 ] ) and ( [ eq : deltapdefavg ] ) of @xmath163 and @xmath108 yield j()=|c|^-1 ( _ q+ _ q ) j(f__i_0 , , _ i_),[eq : jprep ] and _ p , n , k,_1=|c|^-1 ( _ q+ _ q ) _ p , n , k , ( ) . [ eq : dprep ] for correlation hub screening ( @xmath227 ) @xmath374 for all @xmath375 while , as the set @xmath376 s are i.i.d .
uniform for @xmath377 , @xmath378 for @xmath377 .
as @xmath379 is bounded and @xmath380 the relations ( [ eq : prep2 ] ) and ( [ eq : prep3 ] ) are established for the case of correlation screening . for partial correlation hub screening ( @xmath356 )
then , as @xmath381^{-1 } \mu$ ] , the joint densities of @xmath69 and @xmath68 are related by @xmath382 .
therefore , over the range @xmath383 , the @xmath163 and @xmath200 summands in ( [ eq : jprep ] ) and ( [ eq : dprep ] ) are of order @xmath384 and @xmath385 , respectively , which establishes ( [ eq : prep2 ] ) and ( [ eq : prep3 ] ) for partial correlation screening .
this completes the proof of prop .
[ prop : prep ] .
|
one of the most important problems in large scale inference problems is the identification of variables that are highly dependent on several other variables .
when dependency is measured by partial correlations these variables identify those rows of the partial correlation matrix that have several entries with magnitude close to one ; i.e. , hubs in the associated partial correlation graph .
this paper develops theory and algorithms for discovering such hubs from a few observations of these variables .
we introduce a hub screening framework in which the user specifies both a minimum ( partial ) correlation @xmath0 and a minimum degree @xmath1 to screen the vertices .
the choice of @xmath0 and @xmath1 can be guided by our mathematical expressions for the phase transition correlation threshold @xmath2 governing the average number of discoveries .
we also give asymptotic expressions for familywise discovery rates under the assumption of large @xmath3 , fixed number @xmath4 of multivariate samples , and weak dependence . under the null hypothesis that the covariance matrix is sparse
these limiting expressions can be used to enforce fwer constraints and to rank these discoveries in order of statistical significance ( p - value ) . for @xmath5
the computational complexity of the proposed partial correlation screening method is low and is therefore highly scalable .
thus it can be applied to significantly larger problems than previous approaches .
the theory is applied to discovering hubs in a high dimensional gene microarray dataset .
* keywords * gaussian graphical models , correlation networks , nearest neighbor dependency , node degree and connectivity , asymptotic poisson limits , discovery rate phase transitions , p - value trajectories
|
a proper quantification of the joint distribution allowing for the time - varying dependence between assets is critical for asset pricing , portfolio allocation and risk reduction . for a number of years
, finance literature has been studying the risk reduction benefit from international diversification .
after the recent 2008 financial crisis , many researchers have documented possible reduction of these benefits due to rising dependence between markets .
the literature concentrating on the central european markets has been limited though , as it is widely believed that after the enlargement of the european union , these markets became integrated with very limited opportunities for diversification . in this paper , we revisit this line of research , and study the possible benefits from diversification between the czech px and the german dax stock market indices using data covering the five years crisis period . while it is reasonable to believe that the czech and german stock markets show large degree of dependence due to integration of the czech republic into the euro area as well as large dependence of the czech economy on the german one , we aim to study , whether the german and czech stock indices can be considered for the reduction of risk of an international investor .
a number of researchers have addressed the issue of central and eastern european ( cee ) markets integration with the euro area .
@xcite documents increasing stock integration between central european ( ce ) markets and their mature counterparts in europe , and finds lower diversification opportunities at the aggregate stock index level .
@xcite finds long - run cointegrating relationship and hence limited diversification opportunities between cee markets and germany .
@xcite confirm these findings using more recent data .
@xcite analyze the intraday interdependence of western european and central and eastern european markets using wide range of econometric techniques and find evidence of only short - term relationships among the cee and western european stock markets . in more recent study , @xcite
analyze the comovements of three developed and three emerging markets using dcc - garch model on high frequency data .
they detect small correlation between the developed and emerging markets .
this finding is important for investors , as it allows them to diversify their portfolios by investing in the emerging markets .
however , as the authors stress , this diversification opportunity may not be available more recently due to economic integration with western europe .
a study by @xcite use high frequency data to study the foreign macroeconomic announcements and spillover effects on emerging cee stock markets during 2004 - 2007 . among other findings
it is of interest that frankfurt stock market dominates the spillovers over the three emerging markets , while the reaction to the new york market is smaller .
@xcite study the time - varying conditional correlations among the us , germany , russia , and cee markets .
the authors employ the dcc - garch model for correlations and use weekly data spanning from 1997 - 2009 and they find significant increase of correlation between the us and german markets and the cee markets , especially during 2007 - 2009 financial crisis . in a recent work , @xcite
analyze the comovements among western europe vis - - vis central and south eastern europe ( see ) .
the analysis is carried on daily data for the period 2006 - 2011 using bivariate bekk - garch model .
authors find the higher integration among ce and lower integration , with almost zero correlations , among see countries . in another study , @xcite
employ asymmetric dcc - garch to study the comovements in ce markets . using daily data spanning from 2001 - 2011
authors find increase in correlations after these countries joined the european union , whereas asymmetric correlation effects were found only for hungary ( bux ) and poland ( wig ) pair .
in addition , positive relation among conditional correlations and conditional variances suggesting lower diversifications in turbulent times is confirmed .
a common feature of these studies is that they use cointegration , or multivariate garch to study the dependence , and with some exceptions , they use the data before the 2008 financial crisis .
we contribute to the literature by using a very different approach proposed recently by @xcite , allowing us to model the time - varying joint distribution of stock market returns .
using recently proposed time - varying copula methodology and utilizing high frequency data , we build an empirical model which allows us to study the time - varying benefits from diversification .
in addition , we contribute to the understanding of the relationship using recent data covering crisis years . using the fresh data , and state of the art methodologies ,
we revisit the literature and uncover significant time - varying nature of the benefits from diversification between the px and dax markets .
this finding is particularly interesting as previous literature generally reports decreasing potential for diversification .
the work is organized as follows .
the second section introduces our empirical model composing from the realized garch and generalised autoregressive score time - varying copulas .
the third section introduces the data we use , while fourth section discusses the in - sample and out - of - sample fits of all model specifications , and chooses the one which best describes the data . finally , the fifth section tests the economic implications of our empirical model .
we first evaluate the quantile forecasts , which are central to risk management , and then study the time - varying diversification benefits implied by our model .
the last section concludes .
we introduce the empirical model used for describing the dependence between the german dax and czech px stock indices .
our modeling strategy utilizes high frequency data to capture the dependence in the margins and recently proposed dynamic copulas to model the dynamic dependence .
final model is thus able to describe the conditional time - varying joint distribution of returns .
the methodology is based on the @xcite s ( 1959 ) theorem extended to conditional distributions by @xcite .
the extended @xcite s theorem allows to decompose a conditional joint distribution into marginal distributions and a time - varying copula .
consider the bivariate stochastic process @xmath0 with @xmath1 , which has a conditional joint distribution @xmath2 and conditional marginal distributions @xmath3 and @xmath4 .
then @xmath5 where @xmath6 is the time - varying conditional copula of @xmath7 containing all information about the dependence between @xmath8 and @xmath9 , and @xmath10 available information set , usually @xmath11 .
due to @xcite s theorem , we are able to construct a dynamic joint distribution @xmath2 by linking together any two marginal distributions @xmath3 and @xmath4 with any copula function .
theoretically , there is limitless number of valid joint distribution functions that can be created by combining different copulas with different margins , making this approach very flexible .
the first step in building an empirical model based on copulas is to model the margins .
since the largest part of the dependence in financial time series is in their variance , majority of researchers use the generalized autoregressive conditional heteroscedasticity ( garch ) approach of @xcite in this step .
we use the latest advances in the literature which improve volatility modelling by adding the realized volatility measure to the garch model .
this approach utilizes high frequency data to help in explaining the latent volatility .
compared with standard garch(1,1 ) model where the conditional variance of @xmath12-th asset , @xmath13 is dependent on its past values @xmath14 and past values of @xmath15 , @xcite propose to utilize realized volatility measure and make @xmath16 dependent on the realized variance as well . in this work ,
we restrict ourselves to the simple log - linear specification of the so - called realized garch(1,1 ) .
for the general framework of realized garch(@xmath17,@xmath18 ) models we suggest to consult @xcite . while it is important to model conditional time - varying mean @xmath19 , we also include the standard autoregressive ( ar ) term into the final modeling strategy .
as we will find later , autoregressive term of order no larger than two is appropriate for the dax and px return series , thus we restrict ourselves to specifying ar(2 ) with log - linear realgarch(1,1 ) model as in @xcite @xmath20 where @xmath21 is the constant mean , @xmath16 conditional variance , which is latent , @xmath22 realized volatility measured from high frequency data , @xmath23 , and @xmath24 leverage function .
for the @xmath22 , we use the high frequency data and compute it as a sum of squared intraday returns @xcite .
innovations @xmath25 are modelled by the flexible skewed-_t _ distribution of @xcite .
this distribution has two shape parameters , a skewness parameter @xmath26 controlling the degree of asymmetry , and a degree of freedom parameter @xmath27 $ ] controlling the thickness of tails .
when @xmath28 , the distribution reduces to the standard student s @xmath29 distribution , and when @xmath30 , it becomes skewed normal distribution , while for @xmath30 and @xmath28 , it becomes @xmath31 .
thus after the time varying dependence in mean and volatility is modeled , we are left with innovations @xmath32 which have a constant conditional distribution with zero mean and unit variance .
then the conditional copula of @xmath33 is equal to the conditional distribution of @xmath34 : . ]
@xmath35 with @xmath36 being copula parameters , and @xmath37'$ ] conditional probability integral transform @xmath38 the notion of time - varying copula models was initially introduced by @xcite . in further literature , @xcite develop a model where the multivariate garch is extended by copula functions to capture the remaining dependence .
recently , @xcite propose a stochastic copula models , which allow parameters to evolve as a latent time series .
another possibility is offered by arch - type models for volatility @xcite and related models for copulas @xcite , which allow the parameters to be some function of lagged observables .
an advantage of the second approach is that it avoids the need to integrate out " the innovation terms driving the latent time series processes . when working with time - varying copula models the driving dynamics of the model is of crucial importance . for our empirical model
, we therefore adopt the generalized autoregressive score ( gas ) model of @xcite , which specifies the time - varying copula parameter ( @xmath39 ) as a function of the lagged copula parameter and a forcing variable that is related to the standardized score of the copula log - likelihood .
this type of dynamics reduces the one - step - ahead prediction error at current observation given the current parameter values of the copula function .
consider a copula with time - varying parameters : @xmath40 often , a copula parameter is required to fall within a specific range e.g. correlation for normal or student s @xmath29 copula is required to fall in between values of -1 and 1 . to ensure this , @xcite suggest to transform copula parameter by an increasing invertible function @xmath41 ( e.g. , logarithmic , logistic , etc . ) to the parameter @xmath42 for a copula with transformed time - varying parameter @xmath43 ,
a gas(1,1 ) model is specified as @xmath44=i(\delta_t).\end{aligned}\ ] ] while this specification for the time - varying parameters is arbitrary , @xcite motivate it in a way that the model nests a variety of popular approaches from conditional variance models to trade durations and counts models . also , the recursion is similar to numerical optimisation algorithms such as the gauss - newton algorithm .
having specified the model , the last step is to choose the copula function used in the application .
the time - varying copulas we use in this work are the rotated gumbel , normal and student s @xmath29 .
in addition we use constant copulas for comparison . to save the space ,
we do not provide functional forms of copula functions used in this work .
these can be found in @xcite .
the final dynamic copula realized garch model defines a dynamic parametric model for the joint distribution .
the joint likelihood is defined as @xmath45 where @xmath46 is vector of all parameters to be estimated , including parameters of the marginal distributions @xmath47 and parameters of the copula , @xmath36 .
the parameters are estimated using a two - step estimation procedure , generally known as multi - stage maximum likelihood ( msml ) estimation , first estimating the marginal distributions and then estimating the copula model conditioning on the estimated marginal distribution parameters .
while this greatly simplifies the estimation , inference on the resulting copula parameter estimates is more difficult than usual as the estimation error from the marginal distribution must be taken into account . in result , msmle is asymptotically less efficient than one - stage mle , however as discussed by @xcite , this loss is not great in many cases .
moreover , bootstrap methodology can be used for statistical inference .
one of the appealing alternatives to a fully parametric model is to estimate univariate distribution non parametrically , for example by using the empirical distribution function .
combination of a nonparametric model for marginal distribution and parametric model for the copula results in a semiparametric copula model , which we use for comparison to its fully parametric counterpart .
forecasts based on a semiparametric estimation where nonparametric marginal distribution is combined with parametric copula function are not common in economic literature thus it is interesting to compare it in our modelling strategy .
for the margins of the semi - parametric models , we use the non - parametric empirical distribution @xmath48 introduced by @xcite , which consists of modelling the marginal distributions by the ( rescaled ) empirical distribution .
@xmath49 in this case , the parameter estimation is conducted via maximizing likelihood @xmath50 as it can be seen , the likelihood reduces in estimating the copula parameters only .
however , we should note that the inference on parameters is more difficult than usual , hence we rely on bootstrap inference as advocated in @xcite .
an important issue when working with copulas is the selection of the best copula from the pool .
several methods and tests have been proposed for selection procedure .
the methods proposed by @xcite are based on distance from empirical copula . @xcite
propose the use of pseudo - likelihood ratio test for selecting semiparametric multivariate copula models .. ] a test on conditional predictive ability ( cpa ) is proposed by @xcite .
this is a robust test which allows to accommodate both , unconditional and conditional objectives .
recently , @xcite have proposed a test for comparing predictive ability of competing copulas .
the test is based on kullback - leibler information criterion ( klic ) and its statistics is a special case of the unconditional version of @xcite .
as our main aim is to use the model for forecasting , out - of - sample performance of models will be tested by cpa , which considers the forecast performance of two competing models conditional on their estimated parameters to be equal under the null hypothesis @xmath51 = 0 \\ h_{a1 } & : e [ \hat{\mathbf{l } } ] > 0 \mbox { and } h_{a2 } : e [ \hat{\mathbf{l } } ] < 0,\end{aligned}\ ] ] where @xmath52 .
other advantages of this test are the possibilities to use it for both nested and non - nested models , and also for comparison of parametric and semiparametric models .
the asymptotic distribution of the test statistic is @xmath31 and we compute the asymptotic variance using hac estimates to correct for possible serial correlation and heteroskedasticity in the differences in log - likelihoods .
the data set consists of the 5-minute prices of the prague px and german dax cash indices over the period january 3 , 2008 until may 31 , 2013 , covering the recent recession .
we synchronise the data using the time stamp matching , and eliminate transactions executed on saturdays , sundays , holidays , december 24 to 26 , and december 31 to january 2 due to low activity on these days , which could lead to estimation bias . hence ,
in our analysis we work with data from 1349 days . for the empirical model , we need two time series , namely daily returns and realized variance to be able to estimate the realized garch model in margins . for this ,
obtain daily returns as a sum of logarithmic intraday returns , hence we work with open - close returns .
realized variance is computed as a sum of squared 5-minute intraday returns .
figure [ fig : prices ] plots the development of prices of the px and dax together with its realized volatility .
note that plot of prices is normalized so we can compare the movements , and for the plot of realized volatility , we use daily volatility annualized according to the convention @xmath53 .
strong time - varying nature of the volatility can be noticed immediately for both px as well as dax indices . realized volatility of
the dax is larger in average when compared to the volatility of the px index .
otherwise the volatility has similar distributional properties for both indices .
.parameter estimates from ar(@xmath54 ) _ log - linear _ realized garch(1,1 ) and benchmark garch(1,1 ) with innovations distributed _ skew - t _ and normal respectively .
_ t_-statistics reported in parentheses . [ cols= " < , > , > , > , > , < , > , > , > , > " , ] before modeling the dependence structure between the px and dax , we need to model their conditional marginal distributions first . considering general ar models up to five lags
, we find ar(2 ) to best capture the time - varying dependence in mean of px , while dax has a constant mean .
these results are in line with previous research @xcite .
table [ tab : condvarmodel ] summarizes the realized - garch(1,1 ) fit for both px and dax .
in addition , benchmark garch(1,1 ) model is fit to the data for comparison .
all the estimated parameters are significantly different from zero . by observing partial log - likelihood @xmath55 as well as information criteria
, we can see that including realized measures into the garch model improves the fits significantly .
this is crucial for copulas , as we need to specify the best possible model in the margins to make sure there is no univariate dependence left .
if a misspecified model is used for the marginal distributions , then the probability integral transforms will not be uniformly distributed and this will result in copula misspecification . for the estimated standardized residuals from the ar(2 )
realized garch(1,1 ) , we consider both parametric and nonparametric distributions as motivated earlier in the text . before specifying a functional form for time - varying copula function , we test for the presence of time - varying dependence using the simple approach based on the arch lm test .
the test statistics is computed from the ols estimate of the covariance matrix and critical values are obtained using _ i.i.d . _ bootstrap ( for detailed information , consult @xcite ) .
computing the test for the time - varying dependence between the dax and px up to @xmath56 lags , we find the joint significance of all coefficients .
thus we can conclude that there is evidence against constant correlation for the dax and px .
@llcccccccc & & & & + & & & & + + & & & @xmath57 & & & @xmath57 & + normal & @xmath58 & 0.6042 & ( 0.0188 ) & * 305.90 * & & 0.6053 & ( 0.0157 ) & * 307.83 * + clayton & @xmath59 & 0.8596 & ( 0.0591 ) & * 221.98 * & & 0.9258 & ( 0.0560 ) & * 232.08 * + rgumb & @xmath59 & 1.5819 & ( 0.0385 ) & * 265.24 * & & 1.6130 & ( 0.0323 ) & * 272.99 * + student s @xmath29 & @xmath58 & 0.5960 & ( 0.0192 ) & & & 0.6076 & ( 0.0139 ) & + & @xmath60 & 0.0100 & ( 0.0224 ) & * 305.53 * & & 0.0100 & ( 0.0042 ) & * 307.43 * + sym .
joe - clayton & @xmath61 & 0.3667 & ( 0.0295 ) & & & 0.3991 & ( 0.0318 ) + & @xmath62 & 0.3514 & ( 0.0366 ) & * 273.96 * & & 0.3552&(0.0356 ) & * 279.76 * + & & & & + + & & & @xmath57 & & & @xmath57 & + @xmath63 & @xmath64 & -0.0466 & ( 0.1245 ) & & & -0.0037 & ( 0.1103 ) & + & @xmath65 & 0.0466 & ( 0.0520 ) & & & 0.0207 & ( 0.0496 ) & + & @xmath66 & 0.9139 & ( 0.2137 ) & * 266.69 * & & 0.9927 & ( 0.2491 ) & * 275.95 * + @xmath67 & @xmath64 & 0.0121 & ( 0.2883 ) & & & 0.0131 & ( 0.3311 ) & + & @xmath65 & 0.0244 & ( 0.0371 ) & & & 0.0274 & ( 0.0377 ) & + & @xmath66 & 0.9911 & ( 0.2064 ) & * 312.28 * & & 0.9907 & ( 0.2198 ) & * 314.54 * + @xmath68 & @xmath64 & 0.1466 & ( 0.2740 ) & & & 0.1159 & ( 0.2481 ) & + & @xmath65 & 0.0662 & ( 0.0379 ) & & & 0.0755 & ( 0.0375 ) & + & @xmath66 & 0.8936 & ( 0.1941 ) & & & 0.9188 & ( 0.1720 ) & + & @xmath69 & 0.0115 & ( 0.0069 ) & * 311.41 * & & 0.0120 & ( 0.0092 ) & * 313.43 * + motivated by this finding , we estimate three time - varying copula functions , namely normal , rotated gumbel and student s @xmath29 using the gas framework described in the methodology part . as a benchmark , we also estimate the constant copulas to be able to compare the time - varying models against the constant ones . while semiparametric approach is empirically interesting and not often used in literature , we use it for all the estimated models as well
. table [ tab : univrgtvdax - px ] shows the fit from all estimated models . starting with constant copulas , all the parameters are significantly different from zero and normal copula seems to describe the dax and px indices best according to the highest log - likelihood .
semiparametric specifications combining nonparametric distribution in margins with parametric copula function bring further improvement in the log - likelihoods .
importantly , time - varying specifications bring large improvement in log - likelihoods and confirm strong time - varying dependence between the dax and px indices . due to the large number of degrees of freedoms ,
@xmath68 copula in fact converges to the normal one @xmath67 , and thus time - varying normal copula again best describes the data .
this is interesting finding , as it confirms that after proper models for the dependence in margins of the distribution , there is no asymmetry left and the px - dax bivariate distribution is standard normal . to study the goodness of fit for all the specified models , we use kolmogorov - smirnov ( ks ) and cramer - von mises ( cvm ) test statistics with @xmath17-values obtained from 1000 simulations .
none of the fully parametric models is rejected , while most of the semiparametric models are rejected with exception of constant student s @xmath29 , sym .
joe - clayton and time - varying student s @xmath29 .
these results suggest that fully parametric models with realized garch and parametric distribution in margins are all well - specified .
thus realized garch seems to very well model all the dependence in margins , which is crucial for the good specification of the model in the copula - based approach .
semiparametric models are interestingly rejected and are not specified well , except for few mentioned cases .
this is in line with results of @xcite , who finds rejections in semiparametric specifications on the u.s .
indices data .
still , both tests strongly support the realized garch time - varying gas copulas for modeling the joint distribution between dax and px . while it is important to have a well - specified model which describes the data , most of the times we are interested in using the model for forecasting .
thus we conduct an out - of - sample evaluation of the proposed models . for this
, the sample is divided into two periods .
the first , in - sample period , is used to obtain parameter estimates from all models and spans from january 3 , 2008 to may 2 , 2012 . the second
, out - of - sample period is then used for evaluation of forecasts . due to highly computationally intensive estimation of the models ,
we restrict ourselves to a fixed window evaluation , where the models are estimated only once , and all the forecasts are done using the recovered parameters from this fixed in - sample period .
this makes it even harder for the models to perform well in the highly dynamic data . for the out - of - sample forecast evaluation
, we use the conditional predictive ability ( cpa ) test of @xcite .
the time - varying copula models outperforms significantly the constant copula models in out - of - sample evaluation .
this holds both for parametric and semiparametric cases .
thus time - varying copulas have much stronger support for forecasting the dynamic distribution of the dax and px . when comparing the different time - varying copula functions , the test is not so conclusive .
while student s @xmath29 and normal statistically outperform rotated gumbel , the forecasts from student s @xmath29 can not be statistically distinguished from the normal copula .
time - varying normal and student s @xmath29 copulas are thus best in the forecasting exercise .
finally , forecasts from parametric models statistically outperform those from semiparametric ones .
thus , the out - of - sample results confirm the in - sample ones , which is a good sign of proper model fit .
the joint distribution of the px and dax indices is best modeled with the ar(2)-realized garch(1,1 ) time - varying normal copula model .
having correctly specified the empirical model capturing the dynamic joint distribution between the dax and px , we can proceed to studying the pair .
figure [ fig : correlations ] plots the time - varying correlations implied by our model with normal gas copula .
the dependence is generally strong , and also has strong time - varying nature during the studied period .
during the last quarter of the year 2008 , when stock markets were declining due to the lehman brother s crash , the correlation of the px and dax markets rose nearly to 0.7 . in the following year
, it dropped to 0.45 levels and from the year 2010 rose back to 0.7 again .
this result has serious implications for investors as it suggests that diversification possibilities are rapidly changing over past few years during the financial crisis .
we are going to utilize the results and study the possible economic benefits of the modeling strategy .
while it is important to have statistically correct fits , or even good out - of - sample forecasts , the crucial question is whether it translates to economic benefits .
here we test our proposed methodology for economic implications .
first , we quantify the risk of an equally weighted portfolio composed from the dax and px , and second , we study the benefits from diversification to see how the strongly varying correlation affect them .
this is mainly interesting to the international investors considering the czech px stock market index and the german dax index in the portfolio .
quantile forecasts are central to risk management decisions due to a widespread value at risk ( var ) measurement .
var is defined as the maximum expected loss which may be incurred by a portfolio over some horizon with a given probability .
let @xmath70 denote an @xmath71 quantile of a distribution .
var of a given portfolio at time @xmath29 is then simply @xmath72 thus choice of the distribution is crucial to var calculation .
for example assuming normal distribution may lead to underestimation of the var .
our objective is to estimate one - day - ahead var of an equally weighted portfolio composed from the dax and px returns as @xmath73 , which has conditional time - varying joint distribution @xmath74 . in the previous analysis
, we have found that the realized garch model with time - varying normal gas copula well fits and forecasts the data , thus we use it in var forecasts to see whether it correctly forecasts also the joint distribution . as there is no analytical formula , which can be used for this , we rely on monte carlo approach , where we simply simulate the future conditional joint distribution from the estimated models .
[ tab1var ] while quantile forecasts can be readily evaluated by comparing their actual ( estimated ) coverage @xmath75 , against their nominal coverages rate , @xmath76 $ ] , this approach is unconditional and does not capture the possible dependence in the coverage rates .
number of approaches has been proposed for testing the appropriateness of quantiles conditionally , for the best discussion see @xcite .
in our out - of - sample var testing , we use an approach originally proposed by @xcite , who use the @xmath77-th order autoregression @xmath78 , where @xmath79 is @xmath80 if @xmath81 and zero otherwise . while hit sequence @xmath82 is a binary sequence , @xmath83 is assumed to follow a logistic distribution and we can estimate it as a simple logit model and test whether @xmath84 . to obtain the @xmath17-values ,
we rely on simulations as suggested by @xcite and we refer to this test as a @xmath85 test in the results .
moreover , we evaluate the accuracy of var forecasts statistically by defining the expected loss of var forecast made by a forecaster @xmath86 as @xmath87 \left [ y_{t , t+1}-q^{\alpha , m}_{t , t+1 } \right],\ ] ] which has been proposed by @xcite . then , differences in the values of @xmath88 can be tested using @xcite approach , where we test the null hypothesis that the loss function of a benchmark forecaster is the same as the loss function of the tested forecaster @xmath86 , under the alternative that benchmark forecaster is more accurate than the competing one .
table [ tab1var ] reports out - of - sample var evaluation of all models , and figure [ fig : vars ] illustrates the 1% and 5% estimated quantiles of the portfolio .
we can see that all the time - varying models are well specified and the conditional quantile forecasts from them are not rejected by the dq test . for the statistical testing , we use time - varying normal copula as a benchmark forecaster and test all the other models against it .
when looking at the loss functions @xmath89 , we can see that all the quantiles implied by the different models can not be distinguished from each other statistically , except the 1% quantile .
this is mainly because student s @xmath29 copula has large number of degrees of freedom basically converged to the normal one .
thus overall , ar(2)-realized garch(1,1 ) with time - varying copula models are able to describe and forecast the quantiles of the px - dax distribution very well . in case
the dependence of the assets is changing over time strongly , it needs to translate to changing diversification benefits as well .
unlike var , expected shortfall satisfies the sub - additivity property and is a coherent measure of risk .
motivated by these properties , @xcite propose a measure capturing the dynamics in diversification benefits based on expected shortfall .
the conditional diversification benefit ( cdb ) for a given probability level @xmath71 is defined by using time - varying normal copula together with bootstraped confidence band for the constant conditional diversification benefits.,scaledwidth=70.0% ] @xmath90 where @xmath91 is expected shortfall of the portfolio at hand , @xmath92 , \text{for } \alpha \in ( 0,1),\ ] ] @xmath93 is upper bound of the portfolio expected shortfall being the weighted average of the asset s individual expected shortfalls , and @xmath94 lower bound on the expected shortfall being the inverse cumulative distribution function for the portfolio . in other words , this lower bound corresponds to the case where the portfolio never loses more than its @xmath71 distribution quantile .
the measure is designed to stay within @xmath95 $ ] interval , and is increasing in the level of diversification benefits . when the cdb is equal to zero , there are literarily no benefits from diversification , when it equals one , the benefits from diversification are highest possible .
figure [ fig : cdb ] plots the conditional diversification benefits for the px and dax portfolio implied by our empirical model for @xmath96 . similarly to the var case ,
as there is no closed form solution to our empirical model available , we rely on the simulations for cdb computation . encouraged by the previous results , we compute the cdb for the ar(2 ) realized garch with time - varying normal copula model .
analysis could be taken step ahead by optimizing portfolio weights for the highest diversification benefits .
this is done in @xcite , who basically find very small increase , implying that equally weighted portfolio is usually very close to optimal if cdb is used .
also please note that here we do not exploit the full potential of dynamic asset allocation .
diversification benefits vary over time greatly . from the beginning of the sample ,
the benefits from diversification between the dax and the px index are rising gradually until the end of the year 2009 , when they start to decline .
the lowest values are at the beginning of the year 2012 , while from this point until the 2013 , the benefits stay more or less lower .
to support our results , we also report 90% bootstrapped confidence bands computed around a constant level of diversification benefits . assuming the returns data are independently distributed over time with the same unconditional correlation as the px and dax pair , bootstrap confidence level can be conveniently computed via simulations .
we use 10.000 simulations , and report the mean value together with distribution of constant conditional benefits in figure [ fig : cdb ] .
we can see , that the time - varying nature of the conditional diversification benefits is statistically significant , as it departs from the simulated constant distribution .
thus contrary to the general expectation of no diversification benefits for investors considering the czech px index as a diversification tool for the german dax due to very high correlation between these two stocks , we find actual benefits which are varying strongly in time .
this work revisits the czech px and german dax stock markets dependence with the aim to study the opportunities of these two assets in portfolio management . using an empirical model utilizing high frequency data in the time varying copulas , we study the joint conditional distribution of the px and dax returns .
the final ar(2 ) realized garch(1,1 ) with time - varying normal copula is able to capture the dynamics accurately , yielding precise quantile forecasts . using the crisis data , we study the time - varying correlations between the px and dax returns . more important , we study how the time - varying dependence translates to the conditional diversification benefits .
the main result is that the possible diversification benefits are strongly varying over time , and hence even after the 2008 financial crisis , it may be economically interesting to use the dax and px returns for the risk diversification .
this is important finding , as it is contrary to the belief that crisis caused reduction in international diversification benefits .
czech and german economies are strongly tightened as well , so one would expect that especially after the inclusion of the czech republic into the euro area , diversification benefits will disappear .
|
one of the findings of the recent literature is that the 2008 financial crisis caused reduction in international diversification benefits . to fully understand the possible potential from diversification
, we build an empirical model which combines generalised autoregressive score copula functions with high frequency data , and allows us to capture and forecast the conditional time - varying joint distribution of stock returns . using this novel methodology and fresh data covering five years after the crisis
, we compute the conditional diversification benefits to answer the question , whether it is still interesting for an international investor to diversify . as diversification tools , we consider the czech px and the german dax broad stock indices , and we find that the diversification benefits strongly vary over the 20082013 crisis years .
portfolio diversification , dynamic correlations , high frequency data , time - varying copulas c14 , c32 , c51 , f37 , g11
|
in the present paper , we consider the following semilinear stochastic partial differential equation driven by a finite - dimensional wiener process : @xmath0{\mathrm{d}}t+\sigma(u)\,{\mathrm{d}}w,\quad x\in{\mathbb{t}}^n,\;t\in(0,t),\\ u(0)&=u_0 , \end{split}\ ] ] where @xmath1 is a strongly elliptic differential operator , @xmath2 is generally nonlinear unbounded operator and the diffusion coefficient in the stochastic term is also nonlinear .
it is a well known fact in the field of pdes and spdes that many equations do not , in general , have classical or strong solutions . unlike deterministic problems , in the case of stochastic equations we can only ask whether the solution is smooth in the space variable . thus , the aim of the present work is to determine conditions on coefficients and initial data under which there exists a spatially smooth solution to .
the literature devoted to the regularity for linear spdes is quite extensive mainly due to krylov ( see @xcite ) , krylov and rozovskii ( see @xcite , @xcite and the references therein ) and flandoli ( see @xcite )
. however , there seems to be less papers concentrated on regularity for nonlinear spdes .
a class of second order parabolic semilinear spdes was studied by gyngy and rovira ( see @xcite ) but they were only concerned with @xmath3-valued solutions .
so our work can be regarded as an extension of their result .
related problems were also discussed by zhang ( see @xcite , @xcite ) , however , his assumptions are not satisfied in our case .
the main difficulty in the case of semilinear equations lies in the nonlinearities @xmath2 and @xmath4 as , in higher order sobolev spaces , we can not expect the lipschitz condition to be satisfied and hence the fixed point argument can not be applied .
in fact , even the linear growth condition does not hold true in general since the norm of a superposition does not grow linearly with the norm of the inner function ( cf .
proposition [ admissible ] , corollary [ admissible1 ] and remark [ first ] ) . in order to deal with
, we proceed in several steps .
first of all , we consider the equation in @xmath5 and apply the banach fixed point theorem to conclude the existence of an @xmath3-valued mild solution .
next , we study the picard iterations as processes having values in sobolev spaces ( @xmath6 and afterwards @xmath7 ) and find suitable uniform estimates which remain valid also for the limit process . as an immediate consequence of the main result
, we obtain a continuous @xmath8-valued solution . here
, we use the sobolev embedding theorem so the stochastic integration in banach spaces ( see @xcite , @xcite ) , i.e. @xmath7 , allows us to weaken the smoothness assumptions on coefficients .
the paper is organised as follows . in section 2 ,
we review the basic setting and state our main result . in section 3
, we collect important preliminary results related to superposition operators . in the final section ,
these results are applied and the proof of the main theorem is established .
this work was motivated by our research in the field of degenerate parabolic spdes of second order ( see @xcite ) , where smooth solutions of certain approximate nondegenerate problems were needed in order to derive the so - called kinetic formulation and to obtain kinetic solution . nevertheless , since the regularity result of the present paper
is based on properties of strongly elliptic operators , generalization to higher order equations does not cause any additional problems .
let us first introduce the notation which will be used later on . we will consider periodic boundary conditions : @xmath9 where @xmath10 is the @xmath11-dimensional torus .
the sobolev spaces on @xmath10 will be denoted by @xmath12 and by @xmath13 we will denote the space of all functions @xmath14 such that @xmath15 we now give the precise assumptions on each of the terms appearing in the above equation .
we will work on a finite - time interval @xmath16,\,t>0.$ ] the operator @xmath1 is a strongly elliptic differential operator of order @xmath17 with variable coefficients of class @xmath18 .
let us assume , in addition , that @xmath1 is formally symmetric and positive , i.e. we assume that @xmath19 belongs to the resolvent set of @xmath1 .
as an example of this operator let us mention for instance the second order differential operator in divergence form given by @xmath20 where the coefficients @xmath21 are real - valued smooth functions and satify the uniform ellipticity condition , i.e. there exists @xmath22 such that @xmath23 let us now collect basic facts concerning strongly elliptic differential operators satisfying our hypotheses ( for a detailed exposition we refer the reader to @xcite ) .
set @xmath24 .
then the linear unbounded operator @xmath25 in @xmath26 defined by @xmath27 is the infinitesimal generator of a bounded analytic semigroup on @xmath26 .
let us denote this semigroup by @xmath28 .
fractional powers of @xmath29 are well defined and their domains correspond to classical sobolev spaces ( see ( * ? ? ?
* section 10 ) ) , i.e. @xmath30 we will also make use of the following property of analytic semigroups ( see ( * ? ? ? * chapter 2 , theorem 6.13 ) ) : @xmath31 the nonlinearity term @xmath2 is defined as follows : for any @xmath32 @xmath33 where @xmath34 and the functions @xmath35 are smooth enough ( exact assumptions will be given later ) .
let us denote by @xmath36 the vector of functions @xmath37 and denote its length by @xmath38 . throughout this article we fix @xmath39 , a stochastic basis with a complete , right - continuous filtration .
let @xmath40 denote the predictable @xmath4-algebra on @xmath41 $ ] associated with @xmath42 .
for simplicity we will only consider finite - dimensional noise , however , the result can be extended to the infinite - dimensional case .
let @xmath43 be a finite - dimensional hilbert space and let @xmath44 be its orthonormal basis .
the process @xmath45 is a @xmath46-dimensional @xmath47-wiener process in @xmath43 , i.e. it has an expansion of the form @xmath48 , where @xmath49 are mutually independent real - valued standard wiener processes relative to @xmath50 .
the diffusion coefficient @xmath4 is then defined as @xmath51 where the functions @xmath52 satisfy the following linear growth condition @xmath53 since we are going to solve in @xmath26 , for @xmath32 , we need to ensure the existence of the stochastic integral as an @xmath26-valued process .
recall , that @xmath3 spaces , @xmath32 , as well as the sobolev spaces @xmath54 , @xmath55 , belong to a class of the so - called 2-smooth banach spaces , which are well suited for stochastic it integration .
( a detailed construction of stochastic integral for processes with values in 2-smooth banach spaces can be found in @xcite or @xcite . )
let us denote by @xmath56 the space of all @xmath38-radonifying operators from @xmath43 to a 2-smooth banach space @xmath57 .
we will show that @xmath58 for any @xmath59 and @xmath60 note , that the following fact holds true : @xmath61 the proof is , by the way , easy : @xmath62 is an @xmath63-random variable .
let @xmath64 be a sequence of independent @xmath63-random variables , by the definition of a @xmath38-radonifying norm , using and @xmath65 and the claim follows .
the main result of this paper is as follows : [ smooth1 ] let @xmath66 .
we suppose that @xmath67 and @xmath68 have bounded derivatives up to order @xmath69 .
then there exists a solution to which belongs to @xmath70;w^{m , p}({\mathbb{t}}^n)))\ ] ] and the following estimate holds true @xmath71 [ smooth2 ] let @xmath72 and @xmath73 .
assume that @xmath74 have bounded derivatives up to order @xmath75 then there exists a solution to which belongs to @xmath70;c^{k,\lambda}({\mathbb{t}}^n))),\qquad \lambda\in(0,1).\ ] ]
for the reader s convenience we shall first restate the following auxiliary result which is taken from ( * ? ? ? * theorem 5.2.5 ) .
[ admissible ] let @xmath76 .
suppose that the function @xmath77 has bounded derivatives up to order @xmath69 .
if @xmath78 then the following estimate holds true @xmath79 with a constant independent of @xmath36 . since @xmath80 has a linear growth we have @xmath81 next , we will employ the chain rule formula for partial derivatives of compositions : @xmath82 where @xmath83 are multiindices and @xmath84 are certain combinatorial constants .
it is sufficient to consider @xmath85 . by the hlder inequality
we obtain @xmath86 due to interpolation inequalities , we have @xmath87 therefore @xmath88 where we used the fact that @xmath89 is monotone in @xmath90 so the maximal value is attained at @xmath91 or @xmath92 .
the proof is complete .
this result can be easily extended to more general outer function .
[ admissible1 ] let @xmath76 .
suppose that the function @xmath93 has the linear growth latexmath:[\[\label{xx1 } and bounded derivatives up to order @xmath69 .
if @xmath78 then the following estimate holds true @xmath95 with a constant independent of @xmath36 .
[ first ] the situation is much easier for the first order derivatives : fix @xmath96 and let @xmath97 1 .
if @xmath98 with bounded derivative then @xmath99 2 .
if @xmath100 has the linear growth and bounded derivative then @xmath101 where the constant @xmath102 is independent of @xmath36 .
let us review the main ideas of the proof .
the proof is divided into three steps . in the first step ,
we apply the banach fixed point theorem to conclude the existence of an @xmath26-valued mild solution of . in the second step ,
we study picard iterations of and find a uniform estimate of the @xmath103-norm .
it is then used in the third step to derive a uniform estimate of the @xmath12-norm .
this estimate remains valid also for the limit process and the statement follows .
[ prop1 ] let @xmath104 .
assume that @xmath105 and @xmath106 have bounded derivatives of first order .
then there exists a unique mild solution to which belongs to @xmath107,\mathcal{p},{\mathrm{d}}{\mathbb{p}}\otimes{\mathrm{d}}t;l^p({\mathbb{t}}^n)).\ ] ] let us denote @xmath108,\mathcal{p},{\mathrm{d}}{\mathbb{p}}\otimes{\mathrm{d}}t;l^p({\mathbb{t}}^n))\ ] ] and define the mapping @xmath109,\;v\in\mathscr{h}. \end{split}\ ] ] here , we employ stochastic integration in @xmath26 as introduced in section [ notation ] .
we shall prove that @xmath110 maps @xmath111 into @xmath111 and that it is a contraction .
since @xmath105 it follows easily that @xmath112 . in order to estimate the second term , let @xmath113 and note that @xmath114 where the operator @xmath115 commutes with the semigroup and the operator @xmath116 is bounded .
indeed , the operators @xmath117 @xmath118 are clearly bounded . if @xmath119 is the conjugate exponent to @xmath120 and @xmath121 then @xmath122 and the claim follows .
next , all @xmath35 have bounded derivatives hence at most linear growth , so it holds for any @xmath59 @xmath123 indeed , using @xmath120-norm as an equivalent norm on euclidean space @xmath124 @xmath125 if @xmath126 , then using the above remark , the fact , the estimate and the young inequality for convolutions we obtain @xmath127 next , by the burkholder - davis - gundy inequality for martingales with values in 2-smooth banach spaces ( see @xcite , @xcite ) , we have @xmath128 the @xmath38-radonifying norm can be computed using similarly as in .
let @xmath64 be a sequence of independent @xmath63-random variables @xmath129 therefore @xmath130 we conclude that @xmath131 for any @xmath132 . in order to show the contraction property of @xmath110
, we will follow the approach from and use the lipschitz continuity of @xmath36 . indeed , @xmath133 have bounded derivatives so they are lipschitz continuous so @xmath134 can be proved as . for @xmath135 @xmath136 in the case of @xmath137
we employ the same calculations as in and the sequel .
@xmath138 for the @xmath38-radonifying norm we have @xmath139 where the last inequality follows from the fact that all @xmath140 have bounded derivatives therefore are lipschitz continuous .
we conclude @xmath141 consequently @xmath142 where the constant does not depend on @xmath143 and @xmath144 . therefore , if @xmath145 then the mapping @xmath110 has unique fixed point @xmath146 in @xmath111 which is a mild solution of .
the condition on @xmath143 can be easily removed by considering the equation on intervals @xmath147,\,[\tilde{t},2\tilde{t}],\,\dots$ ] with @xmath148 satisfying .
[ prop2 ] let @xmath149 .
assume that @xmath150 and @xmath106 have bounded derivatives of first order .
then the mild solution of belongs to @xmath70;w^{1,p}({\mathbb{t}}^n)))\ ] ] and the following estimate holds true @xmath151 recall that @xmath146 is the limit of picard iterations : let @xmath152 and for @xmath153 define @xmath154 we will show @xmath155 with a constant @xmath102 independent of @xmath156 . by induction on @xmath156 ,
assume that the hypothesis is satisfied for @xmath157 and compute the estimate for @xmath158 .
we will proceed term by term and follow the ideas of proposition [ prop1 ] .
consider the operators @xmath159 restricted to the sobolev space @xmath103 and denote them by @xmath160 .
these operators form a bounded analytic semigroup on @xmath103 generated by the part of @xmath25 in @xmath103 ( see ( * ? ? ?
* theorem v.2.1.3 ) ) .
let us denote this generator by @xmath161 .
therefore we have @xmath162 as above , let @xmath113 and consider the operator @xmath163 we will show that it is a bounded operator . indeed , according to proposition [ prop1 ] , for any @xmath164 , @xmath165 for any multiindex @xmath166 such that @xmath167 , we can write @xmath168 where the operator @xmath169 is bounded . for
each @xmath170 let us fix a multiindex @xmath171 such that it is of order 1 and @xmath172 is also a multiindex , i.e. @xmath173 and @xmath174 .
note , that the operators @xmath175 are bounded as well .
if @xmath119 is the conjugate exponent to @xmath120 we conclude @xmath176 and the claim follows .
therefore , we have @xmath177 to deduce a similar estimate for the stochastic term , we need to consider stochastic integration in @xmath103 .
it holds @xmath178 since @xmath179 , we make use of the maximal estimate for stochastic convolution ( * ? ? ?
* corollary 3.5 ) which can be proved by the factorization method .
for the reader s convenience we recall the basic steps of the proof .
according to the stochastic fubini theorem ( * ? ? ?
* proposition 3.3(v ) ) , @xmath180 where @xmath181 hence application of the hlder , burkholder - davis - gundy and young inequalities yields ( here the constant @xmath102 is independent on @xmath143 )
@xmath182 so @xmath183 and finally @xmath184 where the constant does not depend on @xmath156 .
now , we make use of remark [ first ] and obtain @xmath185 let us make an additional hypothesis : assume that @xmath143 is such that @xmath186 denoting @xmath187 we have @xmath188 and inductively in @xmath156 @xmath189 where @xmath190 is independent @xmath156 .
so follows if @xmath143 is sufficiently small . in order to remove this condition
, we consider a suitable partition of the interval @xmath16 $ ] .
let @xmath191 satisfy and @xmath192 for some @xmath193 .
fix @xmath194 .
we will study the processes @xmath195 on the interval @xmath196 $ ] and find an estimate similar to .
each @xmath195 is the unique mild solution to the corresponding linear equation @xmath197\,{\mathrm{d}}t+\sigma\big(u^{n-1}\big)\,{\mathrm{d}}w,\qquad x\in{\mathbb{t}}^n,\,t\in(0,t),\\ u(0)&=u_0 .
\end{split}\ ] ] let @xmath198 be the mild solution of this problem with the initial condition @xmath144 given at time @xmath199 .
it follows from the uniqueness that for arbitrary @xmath200 @xmath201 and therefore we can write @xmath202 . \end{split}\ ] ] following the same approach as above we obtain @xmath203 with a constant similar to @xmath204 in .
hence @xmath205 where the constant @xmath206 is independent of @xmath207 and @xmath156 .
finally , the estimate follows : @xmath208 we have now all in hand to deduce that the sequence @xmath209 is bounded in @xmath210 and therefore has a weak - star convergent subsequence .
any norm is weakly lower semicontinuous so we get the estimate for the limit process @xmath146 .
moreover , since the stochastic convolution has a continuous modification according to ( * ? ? ?
* corollary 3.5 ) , the proof is complete .
proof of regularity in higher order sobolev spaces ( order greater than 1 ) is more complicated as the norm of a superposition does not , in general , grow linearly with the norm of the inner function ( cf .
proposition [ admissible ] , corollary [ admissible1 ] , remark [ first ] ) .
[ prop3 ] let @xmath211 @xmath212 .
assume that @xmath213 and @xmath68 have bounded derivatives up to order @xmath69 .
then the mild solution of belongs to @xmath70;w^{m , p}({\mathbb{t}}^n)))\ ] ] and the following estimate holds true @xmath214 first , we intend to prove the following estimate for the picard iterations @xmath215 with a constant independent of @xmath156 . by induction on @xmath156 , assume that the hypothesis is satisfied for @xmath157 and compute the estimate for @xmath158 .
the following arguments and calculations are mostly similar to those in proposition [ prop2 ] . recall that according to , we have @xmath216 let us consider the restrictions of the operators @xmath217 to the sobolev space @xmath12 and denote them by @xmath218 . by (
* theorem v.2.1.3 ) , we obtain a strongly continuous semigroup of on @xmath12 generated by part of @xmath25 in @xmath12 .
we denote the generator by @xmath219 .
it follows @xmath220 as above , we employ the following bounded operator : let @xmath113 @xmath221 so @xmath222 and for the stochastic term @xmath223 hence @xmath224 we conclude @xmath225 applying proposition [ admissible ] , corollary [ admissible1 ] and we obtain @xmath226 let @xmath143 satisfy the following condition @xmath227 and define @xmath228 @xmath229 then we have @xmath230 hence inductively in @xmath156 @xmath231 where the constant does not depend on @xmath156 .
therefore follows under the additional hypothesis upon @xmath143 .
however , this condition can be removed by the same approach as in proposition [ prop2 ] . similarly to proposition [ prop2 ] we deduce that the sequence @xmath209 is bounded in @xmath232 and therefore holds true .
existence of a continuous modification follows again from ( * ? ? ?
* corollary 3.5 ) .
let @xmath235 .
according to theorem [ smooth1 ] there exists a solution of which belongs to @xmath70;w^{m , p}({\mathbb{t}}^n))),\qquad\forall p\in[2,\infty).\ ] ] if @xmath236 , then according to the sobolev embedding theorem , the space @xmath12 is continuously embedded in @xmath237 for @xmath238 . hence the assertion follows .
19 h. amann , m. hieber , g. simonett , _ bounded @xmath239-calculus for elliptic operators .
eq . * 7 * ( 1994 ) , 613653
. h. amann , _ linear and quasilinear parabolic problems . _
birkhuser verlag , basel , boston , berlin , 1995 .
z. brzeniak , _ on stochastic convolution in banach spaces and applications .
stoch . rep .
* 61 * ( 1997 ) , 245295 .
z. brzeniak , _ stochastic partial differential equations in m - type 2 banach spaces .
_ potential anal .
* 4 * ( 1995 ) , 145 .
f. flandoli , _ dirichlet boundary value problem for stochastic parabolic equations : compatibility relations and regularity of solutions .
stoch . rep .
* 29 * ( 3 ) ( 1990 ) , 331357 .
i. gyngy , c. rovira , _ on @xmath3-solutions of semilinear stochastic partial differential equations . _ stochastic process
* 90 * ( 2000 ) , 83108 .
m. hofmanov , _ degenerate parabolic stochastic partial differential equations . _ to appear .
n. v. krylov , _ a @xmath240-theory of the dirichlet problem for spdes in general smooth domains .
_ probab . theory related fields * 98 * ( 3 ) ( 1994 ) , 389421 .
n. v. krylov , b. l. rozovskii , _ on the cauchy problem for linear stochastic partial differetial equations .
* 41 * ( 6 ) ( 1977 ) , 1329 - 1347 ; english transl .
* 11 * ( 1977 ) .
n. v. krylov , b. l. rozovskii , _ stochastic evolution equations .
_ itogi nauki i tekhniki .
probl . mat .
* 14 * , viniti , moscow , 1979 , 71146 ; english transl .
j. sov . math .
, * 16 * ( 4 ) ( 1981 ) , 12331277 . m. ondrejt , _ uniqueness for stochastic evolution equations in banach spaces .
_ dissertationes mathematicae * 426 * ( 2004 ) , 163 .
a. pazy , _ semigroups of linear operators and applications to partial differential equation .
_ applied mathematical sciences , vol .
44 , springer - verlag , new york , berlin , heidelberg , tokyo , 1983 .
t. runst , w. sickel , _ sobolev spaces of fractional order , nemytskij operators , and nonlinear partial differential equations .
_ de gruyter series in nonlinear analysis and applications , vol .
3 , walter de gruyter & co. , berlin , 1996 .
x. zhang , _ smooth solutions of non - linear stochastic partial differential equations driven by multiplicative noises . _ sci . china math .
* 53 * ( 2010 ) , 29492972 .
x. zhang , _ regularities for semilinear stochastic partial differential equations . _
249 * ( 2007 ) , 454476 .
|
we study the cauchy problem for a semilinear stochastic partial differential equation driven by a finite - dimensional wiener process .
in particular , under the hypothesis that all the coefficients are sufficiently smooth and have bounded derivatives , we consider the equation in the context of power scale generated by a strongly elliptic differential operator .
application of semigroup arguments then yields the existence of a continuous strong solution .
|
control of fluid flow has been an important area of research and has many practical applications .
the question about controllability of fluid flows has attracted the attention of many researchers , more for incompressible flow but much less for compressible flow . in this paper
we are interested in controllability properties of linearized compressible navier - stokes system . for a compressible ,
isothermal barotropic fluid ( density is a function of only pressure ) , the navier - stokes system in @xmath8 consists of equation of continuity @xmath9 = 0 , \end{array}\ ] ] and the momentum equation @xmath10 \\ [ 3.mm ] \displaystyle = -\nabla p(x , t ) + \mu \bigtriangleup { \bf u}(x , t ) + ( \lambda + \mu ) \nabla [ \mbox{div } { \bf u}(x , t ) ] , \end{array}\ ] ] where @xmath11 is the density of the fluid , @xmath12 denotes the velocity vector in @xmath13 and @xmath14 is an external force field in @xmath15 the pressure satisfies the following constitutive law @xmath16 for some constants @xmath17 , @xmath18 the viscosity coefficients @xmath19 and @xmath20 are assumed to be constant satisfying the following thermodynamic restrictions , @xmath21 , @xmath22 for non - barotropic fluid ( when density is a function of pressure and temperature of the fluid ) , the navier - stokes system consists of the equation of continuity , the momentum equation and an additional thermal energy equation @xmath23(x , t ) + \theta(x , t ) \frac{\partial p}{\partial \theta}(x , t ) \mbox{div } { \bf u}(x , t ) \\ [ 3.mm ] \displaystyle = \kappa \bigtriangleup \theta ( x , t ) + \lambda ( \mbox{div } { \bf u}(x , t))^2 + 2 \mu \sum_{i , j=1}^n \frac{1}{4 } \left(\frac{\partial { \bf u}_i}{\partial x_j } + \frac{\partial { \bf u}_j}{\partial x_i}\right)^2 , \end{array}\ ] ] where @xmath24 denotes the temperature of the fluid , @xmath25 is the specific heat constant and @xmath26 is the heat conductivity constant . for ideal gas , the pressure is given by boyle s law : @xmath27 where @xmath28 is the universal gas constant ( see @xcite ) . in this article
, we first consider the compressible navier - stokes system for non - barotropic fluid in a bounded interval @xmath29 linearized around a constant steady state @xmath30 with @xmath31 and @xmath32 more precisely we consider the system @xmath33 where @xmath34 is the characteristic function of an open subset @xmath35 we choose the following initial and boundary conditions for the system : @xmath36 u(0,t ) = 0 , \quad u(l , t ) = 0 \quad \forall \ \ t > 0 , \\ [ 2.mm ]
\theta(0,t ) = 0 , \quad \theta(l , t ) = 0 \quad \forall \ \ t > 0 . \end{array}\ ] ] in - , @xmath37 , @xmath38 and @xmath39 are distributed controls .
we are interested in the following question : given @xmath40 and @xmath41 can we find interior control functions such that the solution @xmath42 of - satisfies @xmath43 our first main result regarding interior null controllability is the following , [ thm1 ] let @xmath44 i.e. , @xmath45 is a proper subset of @xmath46 let us assume @xmath47 .
the system - is not null controllable in any @xmath48 by the interior controls @xmath49 , @xmath50 and @xmath51 acting on density , velocity and temperature equation respectively .
the above negative result can be extended to the case of less regular interior controls or boundary control .
see remark [ rem : lessregcontrol ] and remark [ rem : bdycontrol ] for more details .
our next positive result shows that , if initial density @xmath52 lies in a more regular space then the linearized system is null controllable by velocity and temperature controls acting everywhere in the domain .
[ thm2 ] let @xmath53 in .
let us denote @xmath54 let us assume @xmath55 .
then for any @xmath56 there exist controls @xmath57 and @xmath58 acting everywhere in the velocity and temperature equation respectively , such that the solution of - satisfies @xmath59 the next result shows that the above result is sharp as null controllability can not be achieved by localized interior velocity and temperature controls . [ thm3 ]
let @xmath53 in .
let @xmath60 i.e. , @xmath61 is a proper subset of @xmath46 let us assume @xmath55 .
the system - is not null controllable in any @xmath48 by the interior controls @xmath62 and @xmath63 acting on velocity and temperature equation respectively .
the proof of these results relies on the observability inequality .
we know that the null controllability of a linear system is equivalent to a certain observability inequality for the solutions of adjoint system ( see @xcite , chapter 2 ) .
to prove the negative results , we will construct particular solutions for the adjoint system such that the observability inequality can not hold . in order
to do that first we will consider the adjoint system in @xmath64 as a terminal value problem .
we will construct highly localized solutions known as `` gaussian beam '' .
similar kind of construction has been used for hyperbolic equations by ralston ( @xcite ) and for wave equations by maci and zuazua ( @xcite ) .
we will prove that solutions are localized in a small neighbourhood of any @xmath65 thus given an observation set , we can always find an interval away from the observation set such that the solutions are localized in that interval . using this
we are able to prove the negative results . to the author s best knowledge these are new results regarding controllability issues of navier - stokes system for non - barotropic fluid . in theorem [ thm3 ]
, we proved a positive result when controls acting everywhere in the equation .
the question then arises : whether positive results could be obtained by using control supported in a small , but moving region , as in @xcite .
rosier and rouchon in @xcite proved that the structurally damped wave equation in one dimension is not null controllable by a boundary control . later on martin , rosier and rouchon in @xcite proved that the same equation in one dimension with periodic boundary conditions , is null controllable with a moving distributed control for sufficiently large time .
chaves - silva , rosier and zuazua in @xcite extend the above result to higher dimension .
the structure of the system considered by the authors in @xcite , in some sense , is similar to the linearized compressible , barotropic navier - stokes system in one dimension as well as in higher dimension .
these issues will be discussed in a future work which is in progress .
we have studied in this paper the null controllability of the linearized compressible navier - stokes system only .
the `` gaussian beam '' construction is used to show negative results .
however one may use other techniques based on the use of nonlinearity ( see @xcite for example ) to achieve controllability results for the full nonlinear system .
there have been some results regarding the control of compressible barotropic fluid models in recent years .
amosova in @xcite considers compressible navier - stokes system for viscous barotropic fluid in one dimension in lagrangian coordinates in a bounded domain @xmath66 with dirichlet boundary condition .
she proves local exact controllability to trajectories for the velocity in any time @xmath56 using a localized interior control on the velocity equation , provided that the initial density is already on the targeted trajectory and initial condition lies in @xmath67 ervedoza , glass , guerrero and puel in @xcite consider the compressible navier - stokes system in one space dimension in a bounded domain @xmath68 .
they prove local exact controllability to constant states @xmath69 with @xmath70 using two boundary controls , both for density and velocity , in time @xmath71 when initial condition lies in @xmath72 chowdhury , ramaswamy and raymond in @xcite consider the compressible barotropic navier - stokes system linearized around a constant steady state @xmath73 with @xmath74 in a bounded domain @xmath75 .
they proved that the linearized system is not null controllable by a localized control or by boundary control .
they also proved that the linearized system is null controllable by an interior control acting everywhere in the velocity equation when initial condition lies in @xmath76 chowdhury in @xcite considers the compressible barotropic navier - stokes system linearized about a constant steady state @xmath77 with @xmath78 in @xmath79 with dirichlet boundary condition and an interior control on the velocity equation acting on open subset @xmath80 he proves that the system is approximately controllable in @xmath81 when @xmath82 he also proves a similar result in two dimension .
chowdhury , mitra , ramaswamy and renardy in @xcite consider the compressible barotropic navier - stokes system linearized about a constant steady state @xmath77 with @xmath78 in @xmath83 with periodic boundary condition .
they proved that the linearized system is null controllable by a localized velocity control when @xmath84 and initial condition lies in @xmath85 our linearized system - is similar to the linearized system considered by the authors in @xcite .
so we expect similar controllability results .
their method is based on explicit expression for eigenfunctions and the behaviour of the spectrum of the linearized operator .
they proved that there is an accumulation point in the spectrum of the linearized operator .
this system behaves very badly with respect to controllability properties and a similar type of controllability behaviour is also observed in @xcite for different types of systems where an accumulation point is present in the spectrum of linearized operator . but the method used in @xcite does not seem to fit very well in our case .
in fact one can prove that there is an accumulation point in the spectrum of the linearized operator considered here , for certain boundary condition .
but the expressions of eigenvalues and eigenfunctions are complicated .
so here we use gaussian beam approach to achieve the negative results .
this technique does not require the knowledge of the spectrum and it seems to extend to higher dimension also .
the controllability properties are completely different , if we consider compressible navier - stokes system linearized around non null velocity .
for barotropic fluid , the system linearized around @xmath86 is not controllable in any time @xmath87 by localized interior control but the system linearized around @xmath88 is null controllable by localized interior control for large time @xmath6 it is interesting to note that there is no accumulation point in the spectrum of the linearized operator in the latter case and better controllability behaviour at least for @xmath87 large enough(see @xcite ) .
but the question remains what happens when time @xmath87 is small enough .
our results answer this question in the negative .
we consider the compressible non - barotropic navier - stokes equation linearized around constant steady state @xmath89
@xmath90 @xmath91 \displaystyle u_t - \frac{\lambda + 2\mu}{{\bar \rho } } u_{xx } + \frac{r { \bar \theta}}{{\bar \rho}}\;\rho_x + \bar v u_x + r \theta_x \ ; = g \chi_{\mathcal{o}_2 } , \mbox { in } ( 0,l ) \times ( 0,t ) , \\[3.mm ] \displaystyle \theta_t ( x , t ) -\frac{\kappa}{{\bar \rho}c_v } \theta_{xx } + \frac{r { \bar \theta}}{c_v } u_x + \bar v \theta_x \ ; = h \chi_{\mathcal{o}_3 } , \mbox { in } ( 0,l ) \times ( 0,t ) , \\[3.mm ] \displaystyle \rho(0 ) = \rho_0,\quad \quad u(0)= u_0 , \mbox { and } \quad \theta(0 ) = \theta_0 , \quad \mbox { in } ( 0,l ) , \\ [ 2.mm ]
\displaystyle \rho(0,t ) = 0 , \;\;\ ; u(0,t ) = 0 = u(l , t ) , \;\ ; \forall \ t \in ( 0,t ) , \\ [ 2.mm ] \displaystyle \theta(0,t ) = 0 = \theta(l , t ) , \;\ ; \forall \ t \in ( 0,t ) .
\end{array}\ ] ] we prove the following theorem .
[ thm1.4 ] let @xmath92 i.e. , @xmath45 is a proper subset of @xmath46 let us assume @xmath47 . if @xmath93 then the system is not null controllable by localized interior controls @xmath94 @xmath50 and @xmath51 acting on density , velocity and temperature equation respectively . as a corollary of the above theorem , one can rule out null controllability of compressible barotropic navier - stokes system linearized around constant steady state @xmath95 in small time @xmath96 using a boundary control or localized interior control .
we consider compressible barotropic navier - stokes system linearized around a constant steady state @xmath95 , @xmath97 in @xmath98 or in @xmath99 as in @xcite . * for initial condition belonging to @xmath100
the system with dirichlet boundary condition is not null controllable at any time @xmath40 by a interior control acting only in the velocity equation .
the control may act in a non empty open subset of @xmath79 or in the whole domain @xmath101 * the system with periodic boundary condition is not null controllable by interior control localized in @xmath102 acting only in the velocity equation when initial condition lies in @xmath103 and time @xmath104 * for initial condition belonging to @xmath100 the same system is not null controllable by boundary control if time @xmath105 from the above corollary we see that the condition @xmath106 in ervedoza , glass , guerrero and puel ( @xcite ) is natural .
next we will show that , our method can be extended to higher dimension also . for simplicity
we consider the compressible barotropic navier - stokes system in two - dimensional bounded domain @xmath107 , linearized around a constant steady state solution @xmath108 , @xmath1 @xmath109 \displaystyle { \bf u } - \frac{\mu}{{\bar \rho } } \delta { \bf u } - \frac{\lambda + \mu}{{\bar \rho}}\nabla[\mathrm{div } \ { \bf u } ]
+ \;a \gamma { \bar \rho}^{\gamma-2 } \nabla \rho \;=\;{\bf g } \chi_{{\mathcal o}_2 } , \mbox { in } \omega \times ( 0,t ) , \\[2.mm ] \displaystyle \rho(0 ) = \rho_0 \quad \mbox{and}\quad { \bf u}(0)= { \bf u}_0 , \quad \mbox { in } \omega , \\ [ 2.mm ]
\displaystyle { \bf u } = { \bf 0 } \ \ \mbox { on } \partial \omega \times ( 0,t ) .
\end{array}\ ] ] where @xmath110 and @xmath111 are open subsets of @xmath112 we obtain the following negative null controllability result for the system . [ thm1.5 ] let @xmath113 i.e. , @xmath45 is a proper open subset of @xmath112 let us assume that @xmath114 then the system is not null controllable in time any @xmath40 by interior controls @xmath115 and @xmath116 the plan of the paper is as follows .
in section 2 , we study the control system linearized around a constant steady state @xmath0 in one dimension .
we prove theorem [ thm1 ] , theorem [ thm2 ] and theorem [ thm3 ] here . in section 3 ,
we study the control system linearized around a constant steady state @xmath117 theorem [ thm1.4 ] is proved here . in section 4
we consider the linearized system in two dimension around constant steady state @xmath7 we prove theorem [ thm1.5 ] here . * acknowledgement * : the author would like to thank prof .
sylvain ervedoza for providing important references on gaussian beams .
the author also would like to thank him and prof .
mythily ramaswamy for very useful discussions which improved the initial version .
the author acknowledges the financial support under the project `` pde control '' from the indo french centre for applied mathematics ( ifcam ) .
in this section , we will discuss interior null controllability of the system - .
we introduce the positive constants @xmath118 let us define @xmath119 endowed with the inner product @xmath120 the following proposition about existence and uniqueness of the system - follows easily from semigroup theory .
let @xmath121 .
let us assume that @xmath115 , + @xmath122 and @xmath123 .
then - has a unique solution @xmath42 with @xmath124 , @xmath125 and @xmath126 .
moreover @xmath127;z).$ ] the idea is to use the adjoint system to derive certain identity which can be used to obtain an observability inequality , equivalent to null controllability .
( see @xcite , chapter 2 ) . for this
we consider the following adjoint problem , @xmath128 \displaystyle - v_t - \nu_0 v_{xx } - \frac{r { \bar \theta}}{{\bar \rho}}\;\sigma_x - r \phi_x \ ; = 0 , \mbox { in } { ( 0,l)}\times ( 0,t ) , \\[3.mm ] \displaystyle - \phi_t - k_0 \phi_{xx } - \frac{r { \bar \theta}}{c_v } v_x \ ; = 0,\ \mbox { in } { ( 0,l)}\times ( 0,t ) , \\ [ 2.mm ]
\displaystyle \sigma(t ) = \sigma_t,\quad v(t ) = v_t , \quad \phi(t ) = \phi_t , \mbox { in } { ( 0,l ) } , \\ [ 2.mm ]
v(0,t ) = 0 = v(l , t ) \quad \forall \ \ t > 0,\quad \phi(0,t ) = 0 = \phi(l , t ) \quad \forall \ \ t > 0 . \end{array}\ ] ] with @xmath129 the adjoint system is also well posed in @xmath130 .
in fact we have let @xmath129 the system has a unique solution with @xmath131 , @xmath132 and @xmath133 .
moreover @xmath134 belongs to @xmath135;z).$ ] let us first assume that @xmath136 @xmath137 + @xmath138 @xmath139 @xmath140 and let @xmath42 and @xmath141 be the solutions of and respectively .
taking inner product in @xmath130 of with @xmath141 and integrating we obtain @xmath142 an integration by parts and use of gives @xmath143 dx + { \bar \rho}^2 \int_0^l [ u(x , t ) v_t(x ) - u_0(x ) v(x,0 ) ] dx \notag \\
+ \frac{{\bar \rho}^2 c_v}{{\bar \theta } } \int_0^l [ \theta(x , t ) \phi_t(x ) - \theta_0(x ) \phi(x,0 ) ] dx = r { \bar \theta}\int_0^t \int_{\mathcal{o}_1 } f \sigma \ ; dx dt \ \notag \\ + { \bar \rho}^2 \int_0^t \int_{\mathcal{o}_2 } g v \ dx dt + \frac{{\bar \rho}^2 c_v}{{\bar \theta } } \int_0^t \int_{\mathcal{o}_2 } h \phi \ dx dt .
\end{aligned}\ ] ] the above relation leads us to the identity equivalent to null controllability . for each initial state @xmath144 , the solution of
the system - can be driven to rest by interior controls @xmath145 , @xmath146 and @xmath147 in time @xmath87 if and only if @xmath148 for all @xmath149 where @xmath134 is the solution of the adjoint system . by a density argument
we deduce that for any @xmath121 and @xmath150 the identity holds .
thus from , it follows that holds if and only if - is null controllable and @xmath151 are the corresponding controls
. one can use the identity to get an observability inequality which is also equivalent to null controllability .
more precisely we have the following proposition ( see @xcite section 2 and @xcite chapter 2 ) .
the system - is null controllable in @xmath130 in time @xmath40 if and only if there exists a constant @xmath152 such that for any terminal condition @xmath153 , @xmath141 , the solution of the adjoint problem satisfies the following observability inequality @xmath154 we now want to prove that the system - is not null controllable in @xmath130 when controls are localized .
our idea is to show that the observability inequality does not hold in this case . for this
we first consider the adjoint problem in whole real line : @xmath155 \displaystyle - v_t - \nu_0 v_{xx } - \frac{r { \bar \theta}}{{\bar \rho}}\;\sigma_x - r \phi_x \ ; = 0 , \mbox { in } \mathbb{r } \times ( 0,t ) , \\[3.mm ] \displaystyle - \phi_t ( x , t ) - k_0 \phi_{xx } - \frac{r { \bar \theta}}{c_v } v_x \ ; = 0 , \mbox { in } \mathbb{r } \times ( 0,t ) , \\ [ 2.mm ]
\displaystyle \sigma(t ) = \sigma_t,\quad v(t ) = v_t , \quad \phi(t ) = \phi_t \quad \mbox { in } \mathbb{r}. \end{array}\ ] ] first we will construct a particular solution of the above adjoint problem which is localized in a small neighbourhood of any @xmath65 for this we would like to have a fourier representation formula for the solution of .
let us assume that @xmath156 and @xmath157 we define @xmath158 the fourier transform of @xmath159 in the space variable @xmath160 for each @xmath161 as follows , @xmath162 we also define @xmath163 and @xmath164 in a similar manner . applying the fourier
transform in , we obtain the following system of ode satisfied by @xmath165 , @xmath166 and @xmath167 @xmath168 \displaystyle - \partial_t { \hat v}(\xi , t)- \nu_0 ( - \xi^2 ) { \hat v}(\xi , t ) - \frac{r { \bar \theta}}{{\bar \rho}}\;(i\xi){\hat \sigma}(\xi , t ) - r ( i\xi){\hat \phi}(\xi , t ) \ ; = 0 , \qquad \xi \in { \mathbb r } , t \in ( 0,t ) , \\[3.mm ] \displaystyle - \partial_t { \hat \phi}(\xi , t ) - k_0 ( - \xi^2){\hat \phi}(\xi , t ) - \frac{r { \bar \theta}}{c_v}(i\xi ) { \hat v}(\xi , t ) \ ; = 0 , \qquad \xi \in { \mathbb r } , t \in ( 0,t ) , \\ [ 2.mm ]
\displaystyle { \hat \sigma}(\xi , t ) = { \hat \sigma}_t(\xi),\quad { \hat v}(\xi , t ) = { \hat v}_t(\xi ) , \quad { \hat \phi}(\xi , t ) = { \hat \phi}_t(\xi ) \quad \xi \in { \mathbb r}. \end{array}\ ] ] let us define @xmath169 then can be written in the following form @xmath170 the unique solution of the above system of ode can be written as @xmath171 we will now discuss some properties of the eigenvalues of @xmath172 let @xmath173 , @xmath174 and @xmath175 be three smooth functions .
let @xmath176 be a cubic polynomial .
let us define the discriminant of the above cubic polynomial @xmath177 now the roots of the above cubic polynomial are given by the following formula @xmath178 where @xmath179 are the three cubic roots of unity , and @xmath180{\frac{d_1(\xi ) + \sqrt{d_1(\xi)^2 - 4 d_0(\xi)^3}}{2 } } \qquad\ ] ] with @xmath181 and when @xmath182 in this formula , @xmath183 and @xmath184{~~}$ ] denote any choice for the square or cube roots , but one has to be consistent with the choice for all @xmath185 .
if @xmath186 and @xmath187 for some @xmath188 the sign of @xmath189 has to be chosen to have @xmath190 i.e. one should define @xmath191 in this case the roots are given by @xmath192{d_1(\xi)}\right ) , \mbox { for } k=1,2,3.\ ] ] if @xmath193 and @xmath187 for some @xmath188 the three roots are equal @xmath194 if @xmath193 and @xmath195 for some @xmath188 there is a double root @xmath196 and a simple root @xmath197 as @xmath198 @xmath199 and @xmath200 are differentiable functions of @xmath188 it is easy to deduce that real part and the complex part of the roots are also differentiable for all @xmath201 we have the following lemma about the properties of the eigenvalues of @xmath172 [ lem:2.5 ] the eigenvalues of @xmath202 always have non positive real part for all @xmath203 let @xmath204 , be the eigenvalues of @xmath202 , where @xmath205 and @xmath206 .
there exists a constant @xmath207 such that for @xmath208 one of the eigenvalues , say @xmath209 satisfies @xmath210 where @xmath211 the eigenvalues of @xmath202 are given by the roots of the characteristic polynomial @xmath212 since it is a polynomial of degree three , it will always have a real root .
now the polynomial @xmath213 is stable i.e. all the roots have negative real part if and only if @xmath214 ( theorem 2.4 , part i of @xcite ) . for our case ,
@xmath215 and it is easy to verify that @xmath216 for all @xmath217 .
so all the eigenvalues have negative real part for all @xmath217 .
for @xmath218 the characteristic polynomial has only one root @xmath219
let @xmath204 , be the eigenvalues of @xmath220 given by the formula .
thus @xmath221 comparing the coefficients of the characteristic polynomial we obtain , @xmath222 it is well known that , if the discriminant @xmath223 then the polynomial has three distinct real roots and if @xmath224 then the equation has one real root and two complex conjugate roots . in our case the discriminant is @xmath225 \xi^{10 } + o(\xi^8).\end{aligned}\ ] ] we also have @xmath226 this leads us to consider the following two cases .
* case i. * let @xmath227 in this case , there exists a positive constant @xmath228 such that @xmath229 and @xmath230 for all @xmath231 hence for @xmath208 @xmath232 and @xmath233 are all real , positive and distinct .
let us define , @xmath234 in , letting @xmath235 we obtain @xmath236 as @xmath237 are all positive , from the first equation of , we deduce that @xmath238 are all bounded .
they are also continuous for @xmath231 as @xmath239 are all distinct and continuous for @xmath208 without loss of generality we assume that @xmath240 from the last equation of , we obtain that @xmath241 converges to @xmath242 as @xmath243 from we obtain , @xmath244 therefore , @xmath245 and @xmath246 both converge as @xmath247 and one of them converges to @xmath248 and the other one to @xmath249 without loss of generality we assume that @xmath250 therefore @xmath251 from the last two equations of , we have @xmath252 letting @xmath235 we deduce that @xmath253 * case ii . * let @xmath254 in this case , @xmath255 thus there exists a constant @xmath228 such that @xmath256 and @xmath195 for all @xmath257 hence for all @xmath258 we have one real root and two complex conjugate roots .
let @xmath259 hence , for all @xmath208 @xmath260 and @xmath209 are all real and positive .
let us define , @xmath261 from we obtain @xmath262 as @xmath263 are all positive , from the first two equations of , we deduce that @xmath264 @xmath265 are all bounded for @xmath266 .
they are also continuous . from the last equation of
, we obtain @xmath267 multiplying the second equation of by @xmath268 and letting @xmath247 we obtain @xmath269 = 0.\ ] ] again multiplying the second equation of by @xmath270 and letting @xmath247 we obtain @xmath271 = 0.\ ] ] now we can proceed as in case i , to obtain @xmath272 in fact more can be said about @xmath273 we have [ lem:2.6 ] let @xmath202 be defined as in and @xmath274 be the eigenvalue satisfying @xmath275 then there exists a constant @xmath276 such that for all @xmath277 @xmath278 further for @xmath208 where @xmath279 as in lemma [ lem:2.5 ] , @xmath209 is differentiable and we have @xmath280 for some positive constant @xmath152 . from lemma [ lem:2.5 ]
, @xmath209 lies on the right half side of the complex plane and it is bounded for all @xmath203 thus there exists a constant @xmath281 such that @xmath278 the coefficients of the characteristic polynomial are differentiable and from the formula it is easy to deduce that , the real and complex parts of the roots are differentiable for @xmath231 for @xmath282 @xmath209 is real and hence differentiable . as @xmath274 is a root of the characteristic equation , we have @xmath283 differentiating the equation with respect to @xmath185 and using the fact that @xmath209 is bounded for all @xmath185 we obtain the estimate . [
lem : eigenfunction ] let @xmath202 be defined as in and @xmath274 is the eigenvalue satisfying , @xmath275 the eigenfunction of @xmath202 corresponding to @xmath274 is @xmath284 where + @xmath285 for @xmath286 sufficiently large , we have @xmath287 for some positive constant @xmath152 . from ,
we first obtain @xmath288 hence for @xmath286 sufficiently large , we obtain @xmath289 and the estimate of @xmath290 follows .
we now want to give a representation formula for solution of .
we have the following proposition .
let us consider @xmath291 for a suitable @xmath292 such that @xmath293 then @xmath294 is the solution of , corresponding to the terminal condition @xmath295 denote @xmath296 to be the solution of corresponding to the terminal condition defined in .
then from , the solution of can be written as @xmath297 it is easy to verify that @xmath298 by taking the inverse of fourier transform , we obtain , @xmath141 defined as in , is the solution of system corresponding to the terminal condition .
we are now interested in the construction of some particular solutions of the equation , which are localized in a neighbourhood of some @xmath299 .
for that , we choose @xmath300 in using a suitable cut off function @xmath301 and analyze the solution @xmath302 of given by .
[ thm4 ] let @xmath301 be a smooth function compactly supported in @xmath303 and of unit @xmath304 norm .
for any @xmath305 , sufficiently small and for any @xmath306 , let us take @xmath307 let @xmath308 and @xmath309 be as in and respectively .
then they satisfy the following * @xmath310 , * for any @xmath311 , there exists a constant @xmath152 independent of @xmath312 such that @xmath313 * for some positive constant @xmath152 independent of @xmath312 , @xmath314 let us denote @xmath315 first we verify that @xmath316 we have @xmath317 thus @xmath318 . now
@xmath319 so we have @xmath320 .
similarly we can show that @xmath321 thus the representation formula is well defined .
now we will prove ( i ) .
note that @xmath322 by parseval s relation we have @xmath323 using lemma [ lem:2.6 ] we have @xmath324 hence we have proved ( i ) . to prove ( ii )
we have using a change of variable formula @xmath325 note that @xmath326 thus for @xmath327 we have , after integration by parts @xmath328 now @xmath329 thus for @xmath312 small , we have for some @xmath152 independent of @xmath312 @xmath330 therefore , for @xmath312 small enough and @xmath331 we have @xmath332 thus there exists a positive constant @xmath333 , such that @xmath334 this proves ( ii ) .
we have @xmath335 similarly we can show that @xmath336 let us give some more properties of @xmath337 which we will need later on to prove the main results . [
lem : improved - regularity ] let @xmath338 and @xmath309 be as in theorem [ thm4 ] .
then @xmath339 and @xmath340 lie in @xmath341 moreover @xmath342 let us first show that @xmath343 @xmath344 thus @xmath343 in a similar way we can show that @xmath345 and @xmath346 belong to @xmath347 next we will show @xmath348 first we have @xmath349 next we have @xmath350 thus we have shown @xmath351 similarly we can show that @xmath352 and @xmath353 also belong to @xmath354 let @xmath355 be as in theorem [ thm4 ] . then for any @xmath356 we have the following estimate @xmath357 first by lemma [ lem : improved - regularity ] , we conclude that @xmath358 proceeding in a similar way as in theorem [ thm4 ] , we first obtain @xmath359 for some positive constant @xmath360 independent of @xmath361 therefore @xmath362 for @xmath312 small .
[ rem : trace ] let @xmath352 and @xmath353 be as in theorem [ thm4 ] .
then @xmath363 for any @xmath364 by lemma , we have @xmath365 and @xmath366 lie in @xmath367 for any @xmath364 we have @xmath368 and @xmath369 thus @xmath370 similarly , we can show @xmath371 now we will use the above construction to prove theorem [ thm1 ] .
first we prove the following theorem .
theorem [ thm1 ] will be a direct consequence of this theorem .
[ thm5 ] let @xmath45 be a proper subset of @xmath79 and @xmath372 then there exists a terminal condition @xmath373 such that the solution of , corresponding to this terminal condition , satisfy the following estimates * @xmath374 + and * @xmath375 as @xmath45 is a proper subset of @xmath376 we choose @xmath377 and @xmath378 such that @xmath379 let us choose @xmath380 is as in theorem [ thm4 ] with the above choice @xmath381 let @xmath382 and @xmath383 are as in and respectively .
let us set @xmath384 let @xmath385 be the restriction of @xmath309 to @xmath98 and @xmath386 be the restriction of @xmath387 to @xmath101 then @xmath388 , @xmath389 and @xmath390 satisfy the following system , @xmath391 \displaystyle - \breve v_t^\epsilon - \nu_0 \breve v^\epsilon_{xx } - \frac{r { \bar \theta}}{{\bar \rho}}\;\breve \sigma^\epsilon_x - r \breve \phi^\epsilon_x \ ; = 0 , \mbox { in } { ( 0,l)}\times ( 0,t),\\[3.mm ] \displaystyle - \breve \phi_t^\epsilon - k_0 \breve \phi^\epsilon_{xx } - \frac{r { \bar \theta}}{c_v } \breve v^\epsilon_x \ ; = 0 , \mbox { in } { ( 0,l)}\times ( 0,t ) , \\ [ 2.mm ]
\displaystyle \breve \sigma^\epsilon(t ) = \breve\sigma^\epsilon_t,\quad \breve v^\epsilon(t ) = \breve v^\epsilon_t , \quad \breve \phi^\epsilon(t ) = \breve \phi^\epsilon_t , \\ [ 2.mm ] \breve v^\epsilon(0,t ) = q^\epsilon_0(t ) , \quad \breve v^\epsilon(l , t ) = q^\epsilon_l(t ) \quad \forall \ \ t > 0 , \\ [ 2.mm ]
\breve\phi^\epsilon(0,t ) = r^\epsilon_0(t ) , \quad \breve\phi^\epsilon(l , t ) = r^\epsilon_l(t ) \quad \forall \ \ t > 0 . \end{array}\ ] ] note that by lemma [ lem : improved - regularity ] and lemma [ rem : trace ] , we have @xmath392 and @xmath393 the above system has a unique solution @xmath394 with @xmath395;z).$ ] they satisfy the following estimates : * from , we have @xmath396 * from theorem [ thm4 ] , @xmath397 * from lemma [ rem : trace ] , @xmath398 let @xmath399 , @xmath400 and @xmath401 satisfy the following system @xmath402 \displaystyle - \widetilde v_t^\epsilon ( x , t)- \nu_0 \widetilde v^\epsilon_{xx } - \frac{r { \bar \theta}}{{\bar \rho}}\;\widetilde\sigma^\epsilon_x - r \widetilde\phi^\epsilon_x \ ; = 0 , \mbox { in } { ( 0,l)}\times ( 0,t ) , \\[3.mm ] \displaystyle - \widetilde\phi_t^\epsilon - k_0 \widetilde\phi^\epsilon_{xx } - \frac{r { \bar \theta}}{c_v } \widetilde v^\epsilon_x \ ; = 0 , \mbox { in } { ( 0,l)}\times ( 0,t ) , \\ [ 2.mm ] \displaystyle \widetilde \sigma^\epsilon(t ) = 0,\quad \widetilde v^\epsilon(t ) = 0 , \quad \widetilde \phi^\epsilon(t ) = 0 , \mbox { in } { ( 0,l ) } , \\ [ 2.mm ]
\widetilde v^\epsilon(0,t ) = - q^\epsilon_0(t ) , \quad \widetilde v^\epsilon(l , t ) = - q^\epsilon_l(t ) \quad \forall \ \ t > 0 , \\ [ 2.mm ] \widetilde \phi^\epsilon(0,t ) = - r^\epsilon_0(t ) , \quad \widetilde \phi^\epsilon(l , t ) = - r^\epsilon_l(t ) \quad \forall \ \ t > 0 . \end{array}\ ] ] as @xmath403 the above system has a unique solution @xmath404;z).$ ] using , we obtain @xmath405 and @xmath406;l^2(0,l ) ) } \leq c \epsilon^{3/4}.\end{aligned}\ ] ] let us set @xmath407 then , @xmath408 satisfy the system and the estimate holds . using the above theorem
now we can prove theorem [ thm1 ] .
* proof of theorem [ thm1 ] .
* theorem [ thm5 ] shows , in the observability inequality , l.h.s @xmath409 and r.h.s @xmath410 hence can not hold .
hence system - is not null controllable .
[ rem : lessregcontrol ] the above result is established when @xmath411 , @xmath412 and @xmath413 .
we can extend this negative result to less regular control .
more precisely we can take @xmath414 and @xmath415 in order to give a sense to the r.h.s . of , we need to replace @xmath416 and @xmath417 in by @xmath418 and @xmath419 respectively .
the observability inequality becomes @xmath420 where @xmath134 is the solution of the adjoint system .
using the same construction as above one can prove that this observability inequality does not hold and hence the system is not null controllable .
[ rem : bdycontrol ] one can use the above gaussian beam construction to rule out null controllability using a boundary control . let us consider the system - with @xmath421 and @xmath422 in this case null controllability is equivalent to the following observability inequality @xmath423 where @xmath134 is the solution of the adjoint system .
we can use the above construction to show that the above observability inequality does not hold and hence the system is not null controllable by a boundary control in any time @xmath40 .
we proved that the system - is not null controllable when initial condition lies in @xmath130 .
so the natural question is if the system is null controllable or not when the initial conditions are regular . in case of barotropic fluid ,
if initial density lies in @xmath424 then the system is null controllable ( see @xcite and @xcite ) . in this section
we choose @xmath55 .
we will first show that - is null controllable by velocity and temperature control only ( i.e. when @xmath53 ) acting everywhere in the domain with this initial regular condition
. then we will show that we can not achieve null controllability by localizing velocity and temperature control .
we consider the following interior control system @xmath425 \displaystyle u_t - \nu_0 u_{xx } + \frac{r { \bar \theta}}{{\bar \rho}}\;\rho_x + r \theta_x \ ; = g \chi_{\mathcal{o}_2 } , \mbox { in } ( 0,l ) \times ( 0,t ) , \\[3.mm ] \displaystyle \theta_t - k_0 \theta_{xx } + \frac{r { \bar \theta}}{c_v } u_x \ ; = h \chi_{\mathcal{o}_3},\ \mbox { in } ( 0,l ) \times ( 0,t ) , \\ [ 2.mm ]
\displaystyle \rho(0 ) = \rho_0,\quad u(0 ) = u_0 , \quad \theta(0 ) = \theta_0 , \mbox { in } ( 0,l ) \\ [ 2.mm ] u(0,t ) = 0 = u(l , t ) \quad \forall \ \ t > 0 , \quad \theta(0,t ) = 0 = \theta(l , t ) \quad \forall \ \ t > 0 . \end{array}\ ] ] here @xmath38 and @xmath39 are velocity and temperature control respectively .
let us first explain why we need average zero condition for initial density .
integrating the density equation of in @xmath79 and using the boundary conditions we deduce that @xmath426 therefore @xmath427 thus if the system is null controllable in time @xmath40 then necessarily @xmath428 let us define @xmath429 we have the following lemma about existence and uniqueness of solution to the system follows easily from semigroup theory .
given @xmath430 @xmath57 and @xmath431 the system has a unique solution @xmath432 with @xmath433 and @xmath434 moreover @xmath42 belongs to @xmath435;v).$ ] * proof of theorem [ thm2 ] .
* let us first take the following system with interior control @xmath436 @xmath437 \displaystyle u_t - \nu_0 u_{xx } + \frac{r { \bar \theta}}{{\bar \rho}}\;\rho_x \ ; = \tilde g , \mbox { in } ( 0,l ) \times ( 0,t ) , \\[3.mm ] \displaystyle \rho(0 ) = \rho_0,\quad u(0 ) = u_0 , \mbox { in } ( 0,l ) , \\ [ 2.mm ]
u(0,t ) = 0 = u(l , t ) \quad \forall \ \ t > 0 . \end{array}\ ] ] by theorem 5.1 of @xcite , we know that for every @xmath438 there exists a control @xmath439 such that the solution of satisfies @xmath440 now we consider the following heat equation @xmath441 \displaystyle \theta(x,0 ) = \theta_0(x ) , x \in { ( 0,l ) } , \ \
\theta(0,t ) = 0 = \theta(l , t ) \quad \forall \ \ t > 0 . \end{array}\ ] ] by theorem 2.66 of @xcite , for every @xmath442 there exists a control @xmath443 such that the solution of satisfies @xmath444 now define @xmath445 where @xmath446 , @xmath447 are the solutions of and respectively .
thus @xmath448 is the solution of the system with @xmath38 and @xmath39 defined as above and it satisfies @xmath449 therefore the theorem follows .
now we want to show that even if @xmath450 null controllability can not be achieved by localized velocity and temperature controls , i.e. , we want to prove theorem [ thm3 ] .
first we will derive an observability inequality . in order to do this
we introduce , @xmath451 differentiating the first equation of @xmath452 we obtain the following system @xmath453 \displaystyle u_t - \nu_0 u_{xx } + \frac{r { \bar \theta}}{{\bar \rho}}\;\alpha + r \theta_x \ ; = g \chi_{\mathcal{o}_2 } , \mbox { in } ( 0,l ) \times ( 0,t ) , \\[3.mm ] \displaystyle \theta_t ( x , t ) - k_0 \theta_{xx } + \frac{r { \bar \theta}}{c_v } u_x \ ; = h \chi_{\mathcal{o}_3 } , \mbox { in } ( 0,l ) \times ( 0,t ) , \\ [ 2.mm ]
\displaystyle \alpha(0 ) = \alpha_0 : = ( \rho_0)_x,\quad u(0 ) = u_0 , \quad \theta(0 ) = \theta_0 , \mbox { in } ( 0,l ) , \\ [ 2.mm ]
u(0,t ) = 0 = u(l , t ) \quad \forall \ \ t > 0 , \quad \theta(0,t ) = 0 = \theta(l , t ) \quad \forall \ \ t > 0 . \end{array}\ ] ] here @xmath454 .
the system is well posed in @xmath130 .
note that , to prove theorem [ thm3 ] , it is enough to show system is not null controllable in @xmath130 by localized interior controls @xmath38 and @xmath39 .
as before we have the following proposition , for every @xmath455 the system is null controllable in time @xmath87 by localized interior controls @xmath456 and @xmath457 if and only if for every @xmath373 the solution of the following adjoint system @xmath458 \displaystyle - v_t - \nu_0 v_{xx } + \frac{r { \bar \theta}}{{\bar \rho}}\;\sigma_{xx } - r \phi_x \ ; = 0 , \mbox { in } ( 0,l ) \times ( 0,t ) , \\[3.mm ] \displaystyle - \phi_t - k_0 \phi_{xx } - \frac{r { \bar \theta}}{c_v } \phi_x \ ; = 0 , \mbox { in } ( 0,l ) \times ( 0,t ) , \\ [ 2.mm ]
\displaystyle \sigma(t ) = \sigma_t , \quad v(t ) = v_t , \quad \phi(t ) = \phi_t , \mbox { in } ( 0,l ) , \\ [ 2.mm ]
\sigma(0,t ) = 0 = \sigma(l , t ) \quad \forall \ \ t > 0 ,
\quad v(0,t ) = 0 = v(l , t ) \quad \forall \ \ t > 0 , \\ [ 2.mm ] \phi(0,t ) = 0 = \phi(l , t ) \quad \forall \ \ t > 0 . \end{array}\ ] ] satisfies
@xmath459 the system is well posed in @xmath130 .
let us consider the adjoint problem in @xmath64 @xmath460 \displaystyle - v_t - \nu_0 v_{xx } + \frac{r { \bar \theta}}{{\bar \rho}}\;\sigma_{xx } - r \phi_x \ ; = 0 , \mbox { in } \mathbb{r } \times ( 0,t ) , \\[3.mm ] \displaystyle - \phi_t - k_0 \phi_{xx } - \frac{r { \bar \theta}}{c_v } \phi_x \ ; = 0 , \mbox { in } \mathbb{r } \times ( 0,t ) , \\ [ 2.mm ]
\displaystyle \sigma(t ) = \sigma_t , \quad v(t ) = v_t , \quad \phi(t ) = \phi_t , \mbox { in } \mathbb{r}. \end{array}\ ] ] applying the fourier transform in , we obtain the following system of ode @xmath461 where @xmath462 eigenvalues of @xmath463 are the same as eigenvalues of @xmath202 .
let @xmath274 be the eigenvalue satisfying @xmath272 the eigenfunction of @xmath463 corresponding to @xmath274 is @xmath464 where @xmath290 is as in lemma [ lem : eigenfunction ] .
thus if we choose @xmath465 the solution of can be written in the following way @xmath466 we have the following theorem .
[ thm8 ] let @xmath467 be as in theorem [ thm4 ] and @xmath309 be as in .
they satisfy the following * @xmath310 * for any @xmath311 , there exists a constant @xmath152 independent of @xmath312 such that @xmath468 * @xmath469 + for some positive constant @xmath152 independent of @xmath312 .
the proof is similar to theorem [ thm4 ] .
now we can proceed in a similar way as in theorem [ thm5 ] to prove theorem [ thm3 ] . in this section
we will discuss null controllability of system .
we want to prove theorem [ thm1.4 ] .
first we have the following proposition about existence and uniqueness of the system .
let @xmath121 .
let us assume that @xmath115 , + @xmath122 and @xmath470 then has a unique solution @xmath42 with @xmath124 , @xmath125 and @xmath471 moreover @xmath42 belongs to @xmath435;z).$ ] proceeding as before we have the following proposition about the equivalence of null controllability and observability inequality . for every initial state @xmath41
the system is null controllable in any time @xmath87 by localized interior controls @xmath94 @xmath456 and @xmath457 if and only if for every @xmath472 @xmath473 the solution of the following adjoint system @xmath474 \displaystyle - v_t ( x , t)- \nu_0 v_{xx } - \frac{r { \bar \theta}}{{\bar \rho}}\;\sigma_x - \bar v v_x - r \phi_x \ ; = 0 , \mbox { in } { ( 0,l)}\times ( 0,t ) , \\[3.mm ] \displaystyle - \phi_t -\frac{k}{{\bar \rho}c_v } \phi_{xx } - \frac{r { \bar \theta}}{c_v } v_x - \bar v \phi_x(x , t ) \ ; = 0 , \mbox { in } { ( 0,l)}\times ( 0,t ) , \\[3.mm ] \displaystyle \sigma(t ) = \sigma_t,\quad \quad v(t)= v_t , \mbox { and } \phi(t ) = \phi_t , \qquad \mbox { in } { ( 0,l)}\times ( 0,t ) , , \\ [ 2.mm ] \displaystyle \sigma(l , t ) = 0 , \;\;\ ;
v(0,t ) = 0 = v(l , t ) , \;\ ; t \in ( 0,t ) , \\ [ 2.mm ] \displaystyle \phi(0,t ) = 0 = \phi(l , t ) , \;\ ; t \in ( 0,t ) .
\end{array}\ ] ] satisfies the following observability inequality @xmath475 the above adjoint system is well posed in @xmath476 .
we want to show the the observability inequality does not hold for small time @xmath87 .
let us first consider the adjoint problem , in @xmath477 as a terminal value problem only @xmath478 \displaystyle - v_t - \nu_0 v_{xx } - \frac{r { \bar \theta}}{{\bar \rho}}\;\sigma_x - \bar v v_x - r \phi_x \ ; = 0 , \mbox { in } \mathbb{r } \times ( 0,t ) , \\[3.mm ] \displaystyle - \phi_t -\frac{k}{{\bar \rho}c_v } \phi_{xx } - \frac{r { \bar \theta}}{c_v } v_x - \bar v \phi_x \ ; = 0 , \mbox { in } \mathbb{r } \times ( 0,t ) , \\[3.mm ] \displaystyle \sigma(t ) = \sigma_t,\quad \quad v(t)= v_t , \mbox { and } \phi(t ) = \phi_t , \qquad \mbox { in } \mathbb{r}. \end{array}\ ] ] we have the following theorem .
let @xmath338 be as in theorem [ thm4 ] .
let @xmath383 be the solution of with this terminal condition .
then they satisfy the following estimates * @xmath479 * for any @xmath311 , there exists a constant @xmath152 independent of @xmath312 such that @xmath480 where @xmath481 * there exists a constant @xmath360 independent of @xmath482 such that @xmath483 let @xmath134 be a solution of .
let us define the transformed functions for @xmath484 @xmath485 then @xmath486 is a solution of the system .
thus from theorem [ thm4 ] , we easily prove the theorem .
we now give the proof of theorem [ thm1.4 ] .
* proof of theorem [ thm1.4 ] .
* we will proceed in a similar way as in theorem [ thm5 ] . in order to obtain a cotradiction to the observability inequality , we need @xmath487 and does not intersect the set @xmath488 for all @xmath489 in particular @xmath490 must lie inside @xmath79 and does not intersect @xmath491 thus we have to choose @xmath492 and @xmath493 properly so that @xmath490 is either a proper subset of @xmath494 or @xmath495 .
let us first choose @xmath496 and @xmath497 such that @xmath498 .
thus if we want @xmath499 to be a proper subset of @xmath494 , we need @xmath500 this implies @xmath501 similarly if we choose @xmath502 then we need @xmath503 thus when @xmath504 the observability inequality does not hold and hence the system is not null controllable .
in this section , we will prove theorem [ thm1.5 ] .
we will construct highly localized solutions , as we did in one dimension .
let us introduce the following constants @xmath505 the system is well posed in @xmath506 as before we have the equivalence between null controllability and observability inequality for every initial state @xmath507 the system is null controllable by localized interior controls @xmath115 and @xmath508 if and only if for every @xmath509 the solution of the following adjoint system @xmath510 \displaystyle - { \bf v}_t - \mu_0 \delta { \bf v } - \gamma_0 \nabla \mathrm{div } \ { \bf v } - b_1 \nabla \sigma \;=\;0 , \mbox { in } \omega \times ( 0,t ) , \\[2.mm ] \displaystyle \sigma(t ) = \sigma_t\quad \mbox{and}\quad { \bf v}(t)= { \bf v}_t , \quad \mbox { in } \omega , \\ [ 2.mm ] \displaystyle { \bf v } = { \bf 0 } \ \ \mbox { on } \partial \omega \times ( 0,t ) .
\end{array}\ ] ] satisfies the observability inequality @xmath511 here @xmath512 the above adjoint system is well posed in @xmath506 we will now use `` gaussian beam '' construction to show that the observability inequality does not hold .
let us first consider the adjoint problem in @xmath513 as a terminal value problem only .
applying the fourier transformation , we obtain the following system of ode * the eigenvalues of @xmath516 are @xmath517 @xmath518 * for all @xmath519 there exists a constant @xmath520 , such that @xmath521 * there exists @xmath522 such that for all @xmath523 , @xmath524 and @xmath270 are all real and distinct . the eigenvalues
satisfy @xmath525 * for @xmath258 the eigenvalues are differentiable and we have @xmath526 for some positive constant @xmath152 . *
the eigenfunction of @xmath516 corresponding to @xmath270 is @xmath527 let @xmath531 with @xmath532 and @xmath533 let @xmath534 be a smooth function compactly supported in the unit ball and of unit @xmath535 norm .
for any @xmath305 , define @xmath536 and @xmath537 as in .
then they satisfy the following estimates * @xmath538 * for any @xmath311 , there exists a constant @xmath152 independent of @xmath312 such that @xmath539 * there exists a positive constant @xmath152 independent of @xmath312 such that @xmath540 thus for @xmath312 small , we have @xmath544 notice that the constant @xmath152 is independent of @xmath361 therefore for @xmath312 small enough and @xmath545 we have @xmath546 thus there exists a positive constant @xmath333 , such that @xmath334 this proves ( ii ) .
|
in this article , we study the null controllability of linearized compressible navier - stokes system in one and two dimension .
we first study the one - dimensional compressible navier - stokes system for non - barotropic fluid linearized around a constant steady state .
we prove that the linearized system around @xmath0 , with @xmath1 @xmath2 is not null controllable by localized interior control or by boundary control .
but the system is null controllable by interior controls acting everywhere in the velocity and temperature equation for regular initial condition .
we also prove that the the one - dimensional compressible navier - stokes system for non - barotropic fluid linearized around a constant steady state @xmath3 , with @xmath1 @xmath4 @xmath5 is not null controllable by localized interior control or by boundary control for small time @xmath6 next we consider two - dimensional compressible navier - stokes system for barotropic fluid linearized around a constant steady state @xmath7 we prove that this system is also not null controllable by localized interior control . * key words . * linearized compressible navier - stokes system , null controllability , localized interior control , boundary control , gaussian beam .
* ams subject classifications .
* 35q30 , 93c20 , 93b05
|
after the discovery of a @xmath0 higgs boson , phenomenological high - energy physics has entered a new era .
the new particle fits the expectation of the minimal standard model ( sm ) .
this model is thus established as an effective field theory ( eft ) that correctly describes all current particle data ( except for still missing possible particle signals for dark matter and additional cp violation ) .
we know about high - energy scales where the effective theory eventually breaks down the scale of neutrino mass generation , the planck scale but those are far outside the reach of collider physics .
the hierarchy between those scales and the electroweak symmetry breaking scale , combined with the fact that all known elementary particles are weakly interacting , puzzles us due to the apparent fine - tuning in perturbative renormalization .
however , the hierarchy puzzle as such has no phenomenological consequences . in principle , the sm may provide a complete description of all present and future collider data , limited just by our ability to do calculations .
nevertheless , the apparent success of the sm does not imply that we have full control over the spectrum at presently accessible energies , say between 100 gev as the electroweak mass scale and a few tev .
first of all , there is the possibility of extra light weakly interacting particles which escape detection at the lhc
. we will not consider this in the present work but investigate new physics above the mass scale of @xmath1 , @xmath2 , and higgs .
the sm is complete as a renormalizable theory and weakly interacting .
hence , it provides a mechanism for suppressing the impact of new physics on observables .
this fact is generally expressed by the decoupling theorem @xcite : all heavy particles ( heavy compared to the masses of @xmath3 , higgs ) can be integrated out , and their physical effects are suppressed by powers of @xmath4 or @xmath5 , where @xmath6 is the effective energy of the measured elementary interaction , and @xmath7 is the mass scale associated with new physics . the eft approach , which has been widely adopted for precision lhc analyses , encodes this in a lagrangian which contains operators of dimension six and , in some cases , eight or even higher @xcite .
decoupling of scalar particles in the case of two - higgs doublet models ( 2hdm ) has been considered in @xcite , as well as in @xcite . for
a new particle with a mass of @xmath8 , the leading corrections to sm particle properties are at the percent level and below .
this is a challenge for lhc analyses . on the other hand , in scattering processes at the lhc
, the partonic energy @xmath6 can easily enter the @xmath9 range , so direct detection is favored .
various classes of new - physics models with extended fermion and gauge sectors can be excluded up to several @xmath9 . however , the current experimental sensitivity on details of the higgs / nambu - goldstone sector is still marginal .
this is due to the fact that the effective energy available for vector - boson scattering in lhc collisions , for instance , is severely suppressed by steeply falling quark and @xmath10 structure functions . in this paper
we study new physics that is coupled to the higgs / nambu - goldstone sector and manifests itself in scattering processes of @xmath1 , @xmath2 , and higgs particles .
the higgs particle does not occur in the initial state and has its own experimental issues , so we restrict the discussion to nambu - goldstone bosons @xcite , which the nambu - goldstone boson equivalence theorem @xcite relates to longitudinally polarized @xmath1 and @xmath2 bosons .
that is , we investigate processes of the class @xmath11 ( @xmath12 ) , where the initial vector bosons are radiated almost on - shell and collinear off initial energetic quarks in the colliding protons .
vector - boson scattering ( vbs ) as a physical process in hadronic collisions has been observed recently by the atlas and cms collaborations @xcite . the sm prediction has been confirmed , but the initial limits on extra interactions are still rather weak , probing an energy scale close to the pair - production threshold of @xmath13 . with higher energy and better precision becoming available at the lhc , and at future lepton and hadron colliders , data will become much more sensitive to new effects in this sector .
there is no reason to restrict the modelling to weak interactions .
in fact , the initially limited experimental resolution and energy reach encourages us to consider new strong interactions , as such deviations from the sm are experimentally most accessible . for decades
, the theory of vbs processes has been the subject of a vast literature , first in the disguise of the low - energy theorem @xcite , for questions of unitarity @xcite and as a means of phenomenological studies @xcite .
a review of recent work can be found in @xcite .
most of those studies were tailored to the higgs - less case , which is by now excluded . in the presence of a light higgs , in the sm , all vbs processes are perturbative and respect unitarity at all energies
. this situation changes drastically once non - sm interactions are present .
regarding the possible scenarios of new physics affecting vbs , there are no significant restrictions from low - energy data or from the absence of lhc discoveries .
asymptotically , the process is determined by the amplitudes of nambu - goldstone boson scattering , where the initial state contains an even number of nambu - goldstone bosons and thus no half - integer representations of @xmath14 .
any bosonic excitation coupling to this state also has integer @xmath14 quantum numbers and thus can not couple left - handed with right - handed sm fermions . in the limit of exact electroweak symmetry , vbs processes and ordinary sm ( fermionic ) processes
thus probe distinct areas of new physics .
electroweak symmetry breaking mixes those sectors , but the mixing terms are again suppressed by the electroweak scale ( in operators , by additional factors of the higgs doublet ) , and are therefore subleading .
the only important constraint is quantum - mechanical unitarity , which is severely violated in a perturbative calculation if we naively insert the dimension - eight operators of the eft .
we have discussed this fact in detail in ref .
@xcite and proposed a framework of unitarization which allows us to augment the sm in an arbitrary way , while maintaining high - energy unitarity and simultaneously matching the new effects to the low - energy eft .
we will adopt this framework , the _ t - matrix scheme _
, for the concrete models below . extending the work of @xcite , in the present paper we consider a wider class of scenarios beyond the sm and beyond the electroweak mass scale . instead of just extrapolating the eft , which generically leads to asymptotic saturation of amplitudes
, we add new states . the quantum numbers of the new states are chosen such that they retain unsuppressed interactions with the vbs system in the limit of vanishing gauge couplings . as mentioned above
, this implies a certain set of quantum - number assignments and , incidentally , suppresses their couplings to the sm fermion sector .
we may consider strongly coupled states , which we would classify as resonances in analogy with mesons in qcd , or weakly coupled states which we would call new elementary particles .
there is a continuous transition between these extremes , such that we can cover all cases on equal footing .
we defer the discussion of vector resonances to a future publication , since those states mix , after ewsb , with @xmath1 and @xmath2 bosons and thus exhibit a possibly different phenomenology .
this limits the model to four distinct cases , namely scalar and tensor resonances with two different assignments of electroweak quantum numbers , respectively .
we embed these states in an extended eft and match this to the low - energy eft where the resonances are integrated out . for the high - energy limit ,
we apply the t - matrix scheme which keeps the model within unitarity bounds when it eventually becomes strongly interacting at energies above the resonance .
the case of a tensor resonance requires special considerations . while renormalizable weakly interacting theories can not include elementary tensor particles , it is nevertheless possible to set up an effective theory which contains a tensor particle and remains weakly interacting over a considerable range of energies
this has been observed in the context of gravity in extra dimensions @xcite , where massive tensor particles arise in the low - energy effective theory .
massive gravitons provide a very specific pattern of couplings to the higgs doublet , gauge bosons and fermions .
we will set up a more generic model where such relations are absent , and construct a lagrangian description of stckelberg type , where we can separate the genuine tensor resonance with a controlled high - energy behavior from unrelated higher - dimensional operators that become relevant asymptotically .
the massive - graviton model emerges as a special case .
( massive ) higher - spin fields have been discussed e.g. in @xcite . given the observation that new resonances can not necessarily be distinguished from asymptotic saturation
if the resonance energy is high and event rates are low , we may ask the question whether the two cases are distinguishable , i.e. , whether a resonance model yields a different prediction from a eft extrapolation with specific coefficients . we will discuss this issue in an exemplary way for specific parameter sets .
furthermore , the new model allows for weakly coupled resonances that do not leave a significant trace in the low - energy eft , but could nevertheless lead to a visible signal in collider data . to obtain numerical results
, we take the unitarized model , which is originally formulated in the gaugeless limit , re - insert gauge couplings and continue the amplitudes off - shell along the lines of @xcite .
this allows us to set up a model definition for a monte - carlo integrator and event generator , which we use to generate partonic event samples for the lhc , cross sections and physical distributions
. a more detailed elaboration of the calculations can be found in @xcite .
we are going to develop models for the high - energy behavior of scattering amplitudes of sm particles .
this can not be done without precisely stating the assumptions that go into those models , and to cast them into convenient notation and parameterization .
first of all , we assume that the sm is a reasonable low - energy effective theory . that is , a weakly interacting ( lagrangian ) gauge field theory with spontaneous @xmath15 symmetry breaking mediated by a complex higgs doublet , supplemented by the standard sets of quarks and leptons , describes all particle - physics data at and below the electroweak scale to a good approximation . regarding the interactions of fermions and vector bosons ,
this conclusion can be drawn from the impressive success in fitting electroweak and flavour data to the sm .
we can not yet be so sure in the higgs sector proper . while the higgs boson was discovered in accordance with the mass range that the precision analysis of electroweak observables suggests , there is still room for sizable deviations from the sm predictions for its couplings . in particular ,
the higgs self - couplings have not been measured at all .
nevertheless , we will assume that those couplings are close to their sm values , such that deviations can be attributed to higher - dimensional terms in the eft .
future data from lhc and beyond will tell whether this is true . if not , we may generalize our findings to a nonlinearly realized higgs sector .
we have set up our parameterization such that this would cause few modifications in the calculations .
a second assumption regards the low - energy spectrum : we assume that there are no additional light particles , such as higgs singlets or extra doublets , below the ew scale .
if this was not true , it would not invalidate the extended - eft approach , but require the low - energy eft to be revised in order to include extra particles as building blocks .
again , the model extensions discussed here would remain unchanged , but we could expect a richer phenomenology of final states that emerge from couplings to the extra light particles .
we want to describe massive tensor and scalar resonances as extensions of the sm , coupled to the scattering channels accessible in vbs .
we start from the low - energy eft , the sm with higher - dimensional operators included , and add a resonance with appropriate spin and gauge quantum numbers to the lagrangian . requiring the assumed symmetries to be manifest ,
uniquely determines the form of the couplings , again in an eft sense , i.e. as an power series expansion of operators in some inverse mass scale @xmath16 .
it is tempting to identify @xmath16 with the resonance mass @xmath7 .
this would imply arbitrary strong interactions at the mass scale of the resonance .
the form of couplings would be arbitrary since for @xmath17 , there is no viable power expansion , and there are no reliable predictions . while this is a conceivable scenario ,
we rather consider a more economical setup where the resonance at mass @xmath7 can be separated from other effects which are attributed to an even higher scale @xmath16 . as we will show below , it is possible and consistent to choose @xmath18 , both for scalar and tensor states .
@xmath16 is then the appropriate scale for all higher - dimensional operators in the extended eft . in the low - energy eft , integrating out the resonance yields well - defined higher - dimensional couplings suppressed by powers of @xmath7 , which combine with the undetermined @xmath16-suppressed coefficients inherited from the extended eft .
depending on their relative magnitude , we may or may not be able to relate the operator coefficients in the low - energy eft to the resonance couplings of the extended eft .
a new massive spin - zero state might appear as another higgs boson .
indeed , a new higgs singlet @xmath19 can couple to the sm higgs doublet @xmath20 via the renormalizable operators @xmath21}\phi$ ] and @xmath21}\phi^2 $ ] , while a new higgs doublet @xmath22 can couple via @xmath23}^2 $ ] and @xmath24}{\operatorname{tr}\left[{\mathbf{h}}'{}^\dagger{\mathbf{h}}'\right]}$ ] .. ] these terms contribute to higgs mixing and self - interactions , but not directly to vbs . in the eft formalism , the observed higgs boson is the only light scalar by definition , and in the renormalizable part of the lagrangian it saturates the vector - boson couplings .
coupling an extra scalar to vbs then requires two higgs - field derivatives @xmath25 and thus introduces an effective dimension - five operator . in a renormalizable extension of the sm higgs sector , after diagonalization new higgses may eventually appear in vbs processes .
however , we have just noted that in the eft formalism , their couplings are higher - dimensional and thus power - suppressed . this is an incarnation of the higgs decoupling theorem .
renormalizability corresponds to the existence of special trajectories in parameter space , where all irrelevant ( i.e. higher - dimensional ) operators can be removed simultaneously from the lagrangian by a nonlinear field redefinition . without a good reason
_ a priori _ for allowing only points on these trajectories , we consider the renormalizable ( possibly weakly interacting ) case as a special case that is included in the general framework .
this applies , in particular , to higgs sector extensions by singlets and doublets , as long as the extra scalars can be considered heavy in the sense of the eft formalism . for our purposes ,
the phenomenology of generic scalar resonances is then very similar to tensor resonances ( see below ) , namely breaking the renormalizability of the sm and inducing higher - dimensional operators both in the low - energy eft where they are integrated out , and in the high - energy model where they appear explicitly in the phenomenological lagrangian .
we will have to apply a unitarization framework in the energy range at and beyond the resonance .
we now turn to massive spin - two particles , postponing spin - one for later investigations , as stated above .
the physical particle corresponds to an irreducible representation of the rotation group in its own rest frame and thus consists of five component fields , mixing under rotation .
strictly speaking , there is no reason to develop a relativistic field theory for a generic interacting spin - two particle .
if there is no uv completion of the interacting model , it is not possible to construct a complete hilbert space and unitary scattering matrix .
however , for convenience of calculation , it is clearly advantageous to embed the tensor particle in the usual relativistic field - theory context of the eft for the sm .
we therefore introduce extra fields , coupled to currents built from sm fields in a lorentz- and gauge - invariant way , in a lagrangian formalism . for the scalar case ,
this is straightforward since a spin - zero particle is represented by a lorentz scalar field that also has a single component . in the tensor case
, we have to deal with the fact that the appropriate lorentz representation has more than five components . in the rest frame ,
the lorentz symmetry ( or its universal cover @xmath26 ) is kinematically broken down to its @xmath27 subgroup , the universal cover of the rotation symmetry .
the lorentz decuplet decomposes into the irreducible spin states @xmath28 looking at the symmetric rest - frame polarization tensor @xmath29 , the irreducible parts correspond to the components @xmath30 ( traceless ) , @xmath31 , @xmath32 , and @xmath33 ( trace ) , respectively . under the full lorentz group
, @xmath29 is also reducible and decomposes into the traceless and trace parts .
however , in the presence of interactions it is not straightforward to maintain this decomposition for off - shell amplitudes @xcite .
our model setup requires that , on - shell , only the pure spin - two state propagates .
if we represent the resonance by a single field , the tensor - field propagator must reduce to the form @xcite @xmath34 here , @xmath35 sums over a basis of five real - symmetric , mutually orthogonal polarization tensors that satisfy the constraints @xmath36 as long as @xmath37 is an on - shell momentum vector , @xmath38 .
the solution to this problem is unique up to the non - resonant part @xcite , @xmath39 where the projection operator of spin - two can be written in terms of the spin - one projection operator , @xmath40 \notag \\ & \quad \qquad\qquad - \frac{1}{3 } p^{\mu_1\mu_2}(k , m)p^{\nu_1\nu_2}(k , m),\end{aligned}\ ] ] with @xmath41 this propagator , with vanishing non - resonant part , can be obtained from the free fierz - pauli lagrangian @xcite coupled to a tensor source @xmath42 @xmath43 in the classical theory , the lagrangian enforces the conditions @xmath44 this is , in principle , a valid lagrangian description of a tensor resonance .
however , since we have to deal with off - shell amplitudes for an effective theory , it will be useful to investigate the role of various terms in more detail .
returning to the propagator ( [ eq : propagatortensor ] ) , there are momentum factors @xmath45 in different combinations that project out the proper spin - two part on the pole .
going to lower energies , these factors vanish more rapidly than the @xmath46 terms and therefore reduce to operators of higher dimension . beyond the resonance
, they will rise more rapidly and therefore potentially provide the dominant part that enters the unitarization prescription . as discussed above ,
the extra momentum factors in the spin - two propagator represent the mismatch between the @xmath47 little group representation of massive on - shell particles and the full lorentz - group off - shell representations in a relativistic description .
this is in analogy with a massive spin - one boson , which in the relativistic case acquires an extra zero component . in the following ,
we identify the extra degrees of freedom for a propagating spin - two object and separate them for the purpose of power - counting in an actual calculation .
to this end , inspired by the spin - one case , we will use the so - called stckelberg formulation for tensor resonances .
this has been studied in the context of effective field theories for massive gravity @xcite , @xcite and @xcite .
the work along these lines has been nicely reviewed in @xcite .
given an arbitrary symmetric polarization tensor @xmath29 that is not restricted by auxiliary conditions , we can subtract terms constructed from momenta , vector and scalar polarizations @xmath48 and demand that ( i ) the fierz - pauli polarization tensor @xmath49 satisfies the on - shell constraints ( [ eq : epsilon - constraints ] ) , and ( ii ) the vector polarization is transversal @xmath50 .
the resulting vector and scalar polarizations @xmath51 , @xmath52 , @xmath53 can be expressed as contractions of the original @xmath29 , [ eq : vst - polarizations ] @xmath54 formally , this subtraction removes the extra representations in the decomposition ( [ eq : decomposition ] ) .
we note that this prescription naturally extends to off - shell wave functions . for the purpose of calculation
, we can reproduce the effect of the propagator ( [ eq : propagatortensor ] ) if we remove all @xmath45 factors from the tensor - field propagator but add a vector and two scalar fields with their respective propagators . to enforce the on - shell relations ( [ eq : vst - polarizations ] ) for their polarization ( i.e. , wave function ) factors , their interactions must be prescribed by the original tensor interactions . in field theory , such relations can be enforced by demanding a gauge invariance . since the momenta have been banished from the numerators of the propagators this way , the power - counting in the resulting feynman rules will be explicit , in analogy with the t hooft - feynman gauge of a gauge theory .
stckelberg @xcite originally formulated the algorithm that systematically introduces the compensating fields together with the extra gauge invariance in the lagrangian formalism . applying the algorithm to the massive tensor case , we start with the fierz - pauli lagrangian which corresponds to the minimal single - field propagator of the pure spin - two tensor . after removing any explicit constraints from the tensor field ,
we introduce first the stckelberg vector @xmath55 that cancels the @xmath56 components , by the replacement @xmath57 and then cancel the extra unwanted @xmath58 components that this field introduces , together with @xmath59 , by a stckelberg scalar @xmath60 , @xmath61 finally , we introduce another stckelberg scalar @xmath19 for cancelling the trace by @xmath62 this scheme guarantees that the interactions of the new fields in the lagrangian are correctly related to the original interactions of the tensor field .
the resulting lagrangian exhibits the gauge invariances that reflect the redundancy of the stckelberg fields and there is a gauge ( called unitary gauge ) in which all stckelberg fields vanish and the original fierz - pauli lagrangian is recovered .
the new fierz - pauli lagrangian with the additional scalar and vector modes reads @xmath63 the scheme simplifies slightly since both scalars are related to the original tensor , so their interactions are not independent .
we can choose the gauge @xmath64 and arrive at a minimal stckelberg lagrangian @xcite ( adjusted by partial integration and simplified ) , @xmath65 for perturbative calculations we have to fix the gauge up to residual gauge transformations @xmath66 that decouple on - shell , i.e. satisfy the harmonic condition @xmath67 .
to this end , we choose linear gauge conditions , @xmath68 and end with a diagonalized lagrangian , @xmath69 next , we normalize the fields canonically @xmath70 and find the canonical propagators @xmath71 for the resulting unconstrained tensor , vector , and scalar fields , respectively .
as desired , these propagators do not contain any momentum factors .
this fact turns out to be essential for a monte - carlo calculation for physical processes , where all bosons are off - shell in a generic momentum configuration .
given this lengthy derivation , we may ask again whether the stckelberg formulation has any advantage over the original fierz - pauli lagrangian .
algebraically , both are equivalent and result in identical on - shell amplitudes .
this should be viewed in analogy with massive vector bosons , for which the stckelberg approach reproduces the usual reformulation as a spontaneously broken gauge theory .
again , this is mathematically equivalent to the original model , as has been pointed out repeatedly @xcite .
however , once the accessible energy in a process exceeds the resonance mass , there is a conceptual difference . in the gauge - theory version ,
there is no higher - dimensional operator with a @xmath72 coefficient .
any additional effects would come with a new cutoff @xmath73 .
scattering amplitudes are bounded beyond the resonance as long as @xmath16 is considered large . by contrast , in the formulation with massive vector bosons , there are @xmath74 terms in the propagator which _ a priori _ require the inclusion of a whole series of operators with @xmath72 factors .
the model is strongly interacting from the onset and has no predictivity .
if actual data show that interactions are indeed weak , this fact would be interpreted as a fine - tuned cancellation among terms . turning this argument around , _ if _ a vector boson is observed to interact weakly over a significant range of scales above its mass , it is natural to describe it as a gauge boson , which in turn determines the allowed interaction pattern .
analogously , _
if _ we assume that a tensor resonance interacts weakly over a significant range of scales above its mass , it is natural to describe it by the stckelberg approach .
we will therefore adopt the stckelberg lagrangian as the basis of a tensor - eft with a minimum set of free parameters . clearly , we can always add extra interactions with further free parameters .
those interactions take the form of higher - dimensional operators which do not contribute on the resonance .
they describe unrelated new - physics effects .
we now combine the findings of the previous section in order to set up a lagrangian description of the resonances , as an extension of the low - energy eft which already ( implicitly ) includes the complete set of higher - dimensional operators . apart from the lorentz representations as scalar or tensor
, we have to consider the representation of the internal symmetry group .
as we will argue in detail below , we take this as the higgs - sector global symmetry @xmath75 , where only the @xmath76 subgroup is gauged .
@xmath77 breaking terms can be systematically included , but we do not consider those in the present work . in the literature on vbs , resonances have traditionally been categorized in terms of weak isospin , i.e. , custodial @xmath78 multiplets .
this is appropriate for a higgsless scenario , where the actual scale of ewsb is given by its natural value @xmath79 ( cf .
e.g. @xcite ) . without a light higgs boson ,
vbs scattering at the lhc would probe the physics at energies below the true ewsb scale , so the ( approximate ) low - energy symmetry applies .
however , since the discovery of the higgs boson , we know that vbs processes probe a scale above the masses of the physical higgs and the electroweak gauge bosons . we have to impose the unbroken high - energy symmetry on the theoretical description . neglecting hypercharge ,
this is @xmath75 .
we therefore describe new resonances coupled to the sm higgs sector in terms of @xmath75 multiplets .
it is not obvious that new physics coupled to the higgs sector actually has this symmetry .
@xmath75 is , first of all , an accidental approximate symmetry of the sm ewsb sector .
there are no possible terms in the dimension - four higgs potential that break @xmath77 , so ewsb leaves the diagonal custodial @xmath27 symmetry untouched
. however , hypercharge and top - quark couplings are not consistent with @xmath77 .
nevertheless , in the gaugeless limit the hypercharge coupling vanishes , and top quarks are irrelevant for vbs anyway , so @xmath77 remains a good symmetry of vbs ( at high @xmath80 ) in the sm . beyond the sm ,
new effects in vbs are transmitted only via the higgs doublet . in the low - energy eft
, they require higher - dimensional operators . these would cause power corrections to the @xmath81 parameter and are therefore constrained by the observed agreement of the measured @xmath81 parameter with the pure sm prediction . for our purposes , we thus adopt @xmath77 as a symmetry of new physics in the higgs sector , to keep things simple .
we have to keep in mind that this need not be the case , and leave the discussion of @xmath77 breaking in this context to future work .
resonances of even spin with unsuppressed couplings to a pair of higgs / nambu - goldstone bosons , must reside in the symmetric part of the decomposition of the product representation of the @xmath82 symmetry , @xmath83 . in the effective interaction operator
, this representation appears as a @xmath84 factor .
there are only two possibilities : 1 .
@xmath85 : a neutral singlet ( isoscalar ) .
@xmath86 : a @xmath87 matrix , which contains nine components .
after ewsb , the multiplet decomposes into an isotensor ( five components ) , an isovector ( three components ) , and an isoscalar ( one component ) . in terms of the gauged @xmath88 subgroup
, the nonet decomposes into a complex @xmath14 triplet with a doubly charged component and a real @xmath14 triplet , as described in @xcite .
the relative mass splitting between these states is of order @xmath89 , where @xmath7 is the average resonance mass . for our purposes where we assume @xmath90
, we ignore that splitting and thus deal with a nonet of degenerate resonance components .
we note that due to the existence of the light higgs , the close analogy between spin and isospin is broken at this point : tensor states have just five physical degrees of freedom , but an isotensor resonance in vbs , given the symmetry assumptions of the present paper , does not exist in isolation .
the distinction comes into play once physical higgs bosons are involved in a process . in vbs amplitudes
, the symmetry relates , for any given resonance multiplet , nambu - goldstone pairs with higgs pairs , i.e. , @xmath91 to @xmath92 production . for simplicity of notation , we will continue to denote the @xmath85 case as isoscalar and the @xmath86 as isotensor , respectively , keeping in mind that the latter case actually is always accompanied by isovector and isoscalar components .
for a scalar isoscalar resonance @xmath60 , we may consider couplings of the form @xmath93}\ ] ] or @xmath94}.\ ] ] the former operator is of lower dimension and might therefore be considered the dominant contribution .
it is part of the higgs potential and influences higgs mixing and production processes . in the present work
, we assume that the scalar state has been broken down in terms of the sm higgs doublet and further states , which themselves arrange as multiplets .
since the sm higgs couplings in the lowest - order eft , the pure sm , saturate the higgs couplings to sm particles and are fixed by definition , residual mixing and potential terms arrange into higher - dimensional operators .
in particular , a resonance coupled to nambu - goldstone bosons is represented by the term ( [ sdhdh ] ) , while the lower - dimensional term ( [ shh ] ) does not enter .
we therefore do not consider ( [ shh ] ) and concentrate on the dimension - five coupling ( [ sdhdh ] ) .
this leads to a current for the scalar isoscalar resonance of the form @xmath95 } \ ; .\end{aligned}\ ] ] while the description of an isoscalar is simple , we have to look at the interactions of the isotensor more carefully . for simplicity
, we will first restrict ourselves to a scalar field multiplet .
a resonance with chiral @xmath75 quantum numbers @xmath86 has nine scalar degrees of freedoms . in the chiral representation
these nine degrees of freedom can be represented as the tensor @xmath96 with the indices @xmath97 .
therefore , the lagrangian describing an isotensor resonance in the nambu - goldstone / higgs boson sector can be written as @xmath98 where the current has a @xmath14 and a @xmath77 index @xmath99}\ , .\end{aligned}\ ] ] analogously to the isoscalar case , the coupling @xmath100 is suppressed by a new physics scale @xmath16 . to expose the coupling structure to the nambu - goldstone / higgs boson sector
, the current can be expanded in the gaugeless limit @xmath101 } = & \frac{1}{2 } \left(\partial_\mu h \partial_\nu h - \partial_\mu w^i \partial_\nu w^i \right ) \delta^{ab } - \frac{1}{2 } \left ( \partial_\mu w^i \partial_\nu h+\partial_\nu w^i \partial_\mu h \right)\varepsilon^{abi }
\\ & + \frac{1}{2 } \left(\partial_\mu w^a \partial_\nu w^b + \partial_\mu w^b \partial_\nu w^a \right ) \end{aligned}\end{aligned}\ ] ] here , the decomposition into isotensor , isovector and isoscalar is already manifest .
the resonance @xmath96 can be represented in a basis constructed from tensor products of @xmath27 generators by the clebsch - gordon decomposition @xmath102 using the basis in the appendix [ appendix : isospin - basis ] , the resonance @xmath96 is rewritten into its @xmath78 components @xmath103 with @xmath104 the lagrangian can be written in terms of the @xmath78 basis [ eq : lagrangian - isotensor - full ] @xmath105 } + { \operatorname{tr}\left[\left(\phi_t + \frac{1}{2}\phi_v - \frac{2}{5}\phi_s \right)j_\phi \right]}\\ j_{\phi } = & f_\phi \left ( \left ( { \mathbf{d}}_\mu { \mathbf{h}}\right ) ^\dagger \otimes { \mathbf{d}}^\mu { \mathbf{h}}+\frac { 1}{8 } { \operatorname{tr}\left[\left ( { \mathbf{d}}_\mu { \mathbf{h}}\right ) ^\dagger { \mathbf{d}}^\mu { \mathbf{h}}\right ] } \right ) ( \tau^a\otimes \tau^a)\end{aligned}\ ] ] in absence of the higgs boson , the coefficient of the second term is chosen in such a way , that the trace of the current vanishes . in this scenario ,
the isovector and isoscalar degree of freedoms decouple from the model and only the isotensor is needed to describe this resonance . however , including a higgs the lagrangian guarantees the amplitude relation between the higgs and nambu - goldstone bosons that will be introduced in section [ section : gaugeless ] .
the crossing relations are manifest in the scattering amplitudes for the nambu - goldstone / higgs boson , which can be determined most easily in the gaugeless limit .
one prominent example for such scalar isotensor resonances appears in the context of composite higgs models of the type little higgs , particularly in the so called littlest higgs model @xcite .
these resonances predominantly couple to the ( electro)weak gauge sector of the sm .
we now construct the effective current that is coupled to a tensor resonance multiplet . by assumption
, the resonance should be produced in vbs processes .
we have to consider independent couplings to the gauge and higgs / nambu - goldstone sectors .
the gauge - sector couplings should vanish in the gaugeless limit , so we are led to consider the higgs - sector coupling .
for a tensor isoscalar resonance , the lowest - dimensional current consists of two terms , @xmath106 } - \frac{c_f}{4 } g^{\mu \nu } { \operatorname{tr}\left [ \left ( { \mathbf{d}}_\rho { \mathbf{h}}\right ) ^\dagger { \mathbf{d}}^\rho { \mathbf{h}}\right ] } \right ) .\end{aligned}\ ] ] the second term actually couples to the trace of the tensor field , which vanishes on - shell .
it is therefore part of the non - resonant continuum and can alternatively be replaced by higher - dimensional operators in the eft .
nevertheless , it is required if , for instance , we want to construct a traceless current .
for now , we leave the coefficient @xmath107 undetermined .
the tensor - field coupling then reads @xmath108 in the fierz - pauli formulation ( section [ sec : fierz - pauli ] ) , and @xmath109 in the stckelberg formulation ( section [ sec : stueckelberg ] ) . in the second version , the momentum factors in the propagator have been turned into derivatives that act on the current .
there is also a coupling to the trace of the current . the formally dominant high - energy ( @xmath110 ) behavior of the amplitude thus is given by the exchange of stckelberg vector and scalar .
the contribution would vanish if the current was conserved .
evaluating the divergence of first and second order , using and in the appendix , @xmath111 } + \frac{f_f}{4 } \left ( c_f + 2 \right ) { \operatorname{tr}\left [ \left ( { \mathbf{d}}_\mu { \mathbf{h}}\right ) ^\dagger\left [ { \mathbf{d}}^\mu,{\mathbf{d}}^\nu \right ] { \mathbf{h}}\right ] } \\ & -\frac{f_f}{4 } \left ( c_f -2 \right ) { \operatorname{tr}\left [ \left ( { \mathbf{d}}_\mu { \mathbf{h}}\right ) ^\dagger\left \ { { \mathbf{d}}^\mu,{\mathbf{d}}^\nu \right \ } { \mathbf{h}}\right ] } \\ = & - f_f \lambda { \operatorname{tr}\left[\widehat{{\mathbf{h}}^\dagger { \mathbf{h}}}\right ] } { \operatorname{tr}\left [ { \mathbf{h}}^\dagger { \mathbf{d}}^\nu { \mathbf{h}}\right ] } \\ & -{\mathrm{i}}g f_f { \operatorname{tr}\left [ \left ( { \mathbf{d}}_\mu { \mathbf{h}}\right ) ^\dagger{\mathbf{w}}^{\mu\nu } { \mathbf{h}}\right ] } -{\mathrm{i}}g^\prime f_f { \operatorname{tr}\left[{\mathbf{h}}{\mathbf{b}}^{\mu\nu } \left({\mathbf{d}}_\mu { \mathbf{h}}\right)^\dagger \right]}\ , , \end{aligned}\end{aligned}\ ] ] @xmath112}\\ & + f_f { \operatorname{tr}\left [ \left({\mathbf{d}}^2 { \mathbf{h}}\right)^\dagger { \mathbf{d}}^2 { \mathbf{h}}\right ] } + f_f { \operatorname{tr}\left [ \left({\mathbf{d}}_\mu { \mathbf{d}}_\nu { \mathbf{h}}\right)^\dagger { \mathbf{d}}^\nu { \mathbf{d}}^\mu { \mathbf{h}}\right]}\\ & - \frac{c_f}{2 } f_f { \operatorname{tr}\left [ \left({\mathbf{d}}_\mu { \mathbf{d}}_\nu { \mathbf{h}}\right)^\dagger { \mathbf{d}}^\mu { \mathbf{d}}^\nu { \mathbf{h}}\right]}\\ = & - f_f \lambda { \operatorname{tr}\left[\widehat{{\mathbf{h}}^\dagger { \mathbf{h}}}\right ] } { \operatorname{tr}\left[{\mathbf{d}}_\mu { \mathbf{h}}^\dagger { \mathbf{d}}^\mu { \mathbf{h}}\right ] } - f_f \lambda { \operatorname{tr}\left[\widehat{{\mathbf{h}}^\dagger { \mathbf{h}}}\right ] } { \operatorname{tr}\left [ { \mathbf{h}}^\dagger { \mathbf{d}}^2 { \mathbf{h}}\right ] } \\ & - 2f_f \lambda { \operatorname{tr}\left[{{\mathbf{h}}^\dagger { \mathbf{d}}_\mu { \mathbf{h}}}\right ] } { \operatorname{tr}\left [ { \mathbf{h}}^\dagger { \mathbf{d}}^\mu { \mathbf{h}}\right]}\\ & + \frac{g^2 f_f}{2 } \left({\operatorname{tr}\left [ \left ( { \mathbf{d}}_\mu { \mathbf{h}}\right ) ^\dagger { \mathbf{h}}\left({\mathbf{d}}^ { \mu } { \mathbf{h}}\right ) ^\dagger{\mathbf{h}}\right ] } -{\operatorname{tr}\left [ \left ( { \mathbf{d}}_\mu { \mathbf{h}}\right ) ^\dagger \left({\mathbf{d}}^ { \mu } { \mathbf{h}}\right ) { \mathbf{h}}^\dagger{\mathbf{h}}\right ] } \right)\\ & + \frac{{g^\prime}^2 f_f}{2 } \left({\operatorname{tr}\left [ \left ( { \mathbf{d}}_\mu { \mathbf{h}}\right ) ^\dagger { \mathbf{h}}\left({\mathbf{d}}^ { \mu } { \mathbf{h}}\right ) ^\dagger{\mathbf{h}}\right ] } -{\operatorname{tr}\left [ \left ( { \mathbf{d}}_\mu { \mathbf{h}}\right ) ^\dagger \left({\mathbf{d}}^ { \mu } { \mathbf{h}}\right ) { \mathbf{h}}^\dagger{\mathbf{h}}\right ] } \right)\\ & + \frac { g^2 f_f}{2 } { \operatorname{tr}\left [ { \mathbf{h}}^\dagger{\mathbf{w}}_{\mu\nu } { \mathbf{w}}^{\mu\nu } { \mathbf{h}}\right ] } + \frac { { g^\prime}^2 f_f}{2 } { \operatorname{tr}\left [ { \mathbf{h}}{\mathbf{b}}_{\mu\nu } { \mathbf{b}}^{\mu\nu } { \mathbf{h}}^\dagger\right ] } \\ & + g{g^\prime } f_f { \operatorname{tr}\left [ { \mathbf{h}}^\dagger { \mathbf{w}}_{\mu\nu } { \mathbf{h}}{\mathbf{b}}^{\mu\nu}\right ] } \\ & -{\mathrm{i}}g f_f { \operatorname{tr}\left [ \left ( { \mathbf{d}}_\mu { \mathbf{h}}\right ) ^\dagger{\mathbf{w}}^{\mu\nu } { \mathbf{d}}_\nu { \mathbf{h}}\right]}\\ & -{\mathrm{i}}g f_f
{ \operatorname{tr}\left [ \left({\mathbf{d}}_\mu { \mathbf{h}}\right ) { \mathbf{b}}^{\mu\nu } \left({\mathbf{d}}_\mu { \mathbf{h}}\right)^\dagger \right ] } \ , , \end{aligned}\end{aligned}\ ] ] we observe that the current is not conserved .
however , none of the nonvanishing terms contributes to the @xmath113 process at high energy .
the stckelberg fields effectively decouple , and the high - energy behavior can be calculated from the propagator ( [ eq : proptensor ] ) .
if we take ewsb into account , we do get a nonvanishing divergence also at the two - particle level .
new terms arise that are proportional to powers of @xmath114 , and thus to the @xmath1 , @xmath2 , and higgs masses .
the stckelberg vector transmits , via ewsb mixing , a coupling to transversal vector bosons . in amplitudes , these factors are accompanied by factors of @xmath115 . in the limit of a heavy resonance , the stckelberg terms are thus parametrically suppressed and become relevant only for energies significantly beyond the resonance mass .
conversely , if the resonance mass is comparable to the electroweak scale , the stckelberg terms are significant .
the remainder of the amplitude that corresponds to the genuine tensor propagator does not contain momentum factors .
nevertheless , the interaction is of dimension five , so we expect contributions that rise with energy .
this occurs for external longitudinally polarized vector bosons which carry a momentum factor .
we obtain a factor @xmath116 in the numerator that asymptotically cancels with the denominator , so the effective rise is proportional to @xmath117 .
qualitatively , this is the same result as for the case of a scalar resonance , or for a higgs - less theory .
we conclude that we can unitarize the amplitude uniformly for all spin - isospin channels , starting from the gaugeless nambu - goldstone boson limit , without having to account for transversal gauge bosons or higher powers of @xmath118 beyond the resonance .
the algorithm can be taken unchanged from the pure - eft case @xcite
. however , we have to restrict the allowed values of resonance masses and couplings such that the stckelberg terms discussed above remain numerically small within some finite energy range . outside this range
, we can no longer separate the higgs / nambu - goldstone sector of the theory but are sensitive to unknown strong interactions that involve all channels of longitudinal , transversal , and higgs exchange simultaneously .
while the unitarization scheme of @xcite is also applicable in that situation , it becomes technically more involved ; we defer this case to future work .
we now list the effective lagrangians that we consider in the subsequent calculations . in all cases ,
the basic theory is the sm eft , i.e. , the sm with the observed light higgs boson in linear representation , extended by higher - dimensional operators .
we add four different resonance multiplets , corresponding to all combinations of spin and isospin 0 and 2 , respectively .
the lagrangians can be combined .
the spin - two lagrangian is presented in the stckelberg gauge . regarding the resonance fields , we should further select electroweak quantum numbers , as discussed in section [ sec : multiplets ] , by defining the precise form of the covariant derivative acting on the resonance field in the kinetic operator . however
, as long as we are not interested in ew radiative corrections , we may work with a simple partial derivative and omit the gauge couplings to @xmath1 , @xmath2 , and photon .
the lagrangian for the isoscalar - scalar @xmath60 , the isotensor - scalar @xmath19 , the isoscalar - tensor @xmath119 and the isotensor - tensor @xmath120 are given by [ eq : appendix_lagrangian ] @xmath121 } + { \operatorname{tr}\left[\left(\phi_t + \frac{1}{2}\phi_v - \frac{2}{5}\phi_s \right)j_\phi \right]}\ , , \\
{ \mathcal{l}}^{}_{f } = & \frac{1}{2 } f_{f\mu\nu}\left ( -\partial^2 - m_f^2 \right ) f_f^ { \mu \nu } + \frac{1}{2 } f^ { \mu}_{f\mu}\left ( -\frac{1}{2 } \left ( -\partial^2 - m_f^2 \right ) \right ) f^{\nu}_{f\nu } \notag \\ & + \frac{1}{2 } a_{f\mu } \left ( -\partial^2 - m_f^2 \right ) a_f^\mu + \frac{1}{2 } \sigma_f \left ( -\partial^2 - m_f^2 \right ) \sigma_f \notag \\ & + \left ( f_{f\mu \nu } - \frac{1}{\sqrt{6 } } g_{\mu \nu }
\sigma + \frac{1}{\sqrt{2}m_f}\left ( \partial_\mu a_\nu + \partial_\nu a_\mu \right ) + \frac{\sqrt{2}}{\sqrt{3}m_f^2 } \partial_\mu \partial_\nu \sigma \right ) j_f^{\mu \nu } \ , , \\ { \mathcal{l}}^{}_{x } = & \frac{1}{2}\sum_{i = s , v , t } \operatorname{tr }
\big [ x^{}_{i\mu\nu}\left ( -\partial^2 - m_x^2 \right ) x_i^ { { } \mu \nu } + x^ { { } \mu}_{i\mu}\left ( -\frac{1}{2 } \left ( -\partial^2 - m_x^2 \right ) \right ) x^ { { } \nu}_{i\nu } \notag \\ & \phantom { \frac{1}{2}\sum_{i = s , v , t } \operatorname{tr } \big [ } + a_{i\mu } \left ( -\partial^2 - m_x^2 \right ) a_i^\mu + \sigma_i \left ( -\partial^2 - m_x^2 \right )
\sigma_i \big ] \notag \\ & + \operatorname{tr } \bigg [ \left ( x^{}_{t\mu \nu } - \frac{g_{\mu \nu}}{\sqrt{6 } } \sigma_t + \frac{\partial_\mu a_{t\nu } + \partial_\nu a_{t\mu } } { \sqrt{2}m_x } + \frac{\sqrt{2}}{\sqrt{3}m_x^2 } \partial_\mu \partial_\nu \sigma_t \right ) j_x^{\mu \nu } \notag \\ & \phantom{+ \operatorname{tr } \big [ } + \frac{1}{2 } { \left ( x^{}_{v\mu \nu } - \frac{g_{\mu \nu}}{\sqrt{6 } } \sigma_v + \frac{\partial_\mu a_{v\nu } + \partial_\nu a_{v\mu } } { \sqrt{2}m_x } + \frac{\sqrt{2}}{\sqrt{3}m_x^2 } \partial_\mu \partial_\nu \sigma_v \right ) j_x^{\mu \nu } } \notag \\ & \phantom{+ \operatorname{tr } \big [ } -\frac{2}{5 } { \left ( x^{}_{s\mu \nu } - \frac{g_{\mu \nu}}{\sqrt{6 } } \sigma_s + \frac{\partial_\mu a_{s\nu } + \partial_\nu a_{s\mu } } { \sqrt{2}m_x } + \frac{\sqrt{2}}{\sqrt{3}m_x^2 } \partial_\mu \partial_\nu \sigma_s \right ) j_x^{\mu \nu } } \bigg ] \ , , \end{aligned}\ ] ] respectively , where the tensor resonances are formulated in the stckelberg formalism with associated fields @xmath122 , @xmath123 and @xmath124 denoting the scalar , vector and tensor degrees of freedom , respectively .
the corresponding stckelberg fields for the isotensor - tensor receive extra indices @xmath125 which represent the isoscalar , isovector and isotensor fields of the @xmath78 multiplet , respectively .
the couplings to the nambu - goldstone boson current in each case is given by @xmath126 } \
, , \\
j_{\phi } & = f_\phi & & \left ( \left ( { \mathbf{d}}_\mu { \mathbf{h}}\right ) ^\dagger \otimes { \mathbf{d}}^\mu { \mathbf{h}}+\frac { 1}{8 } { \operatorname{tr}\left[\left ( { \mathbf{d}}_\mu { \mathbf{h}}\right ) ^\dagger { \mathbf{d}}^\mu { \mathbf{h}}\right ] } \right ) \tau^{aa } \ , , \\ j^{\mu \nu}_f&= f_f & & \left ( { \operatorname{tr}\left [ \left ( { \mathbf{d}}^\mu { \mathbf{h}}\right ) ^\dagger { \mathbf{d}}^\nu { \mathbf{h}}\right ] } - \frac{c_f}{4 } g^{\mu \nu } { \operatorname{tr}\left [ \left ( { \mathbf{d}}_\rho { \mathbf{h}}\right ) ^\dagger { \mathbf{d}}^\rho { \mathbf{h}}\right ] } \right ) \ , , \\ j^{\mu \nu}_{x } & = f_x & & \bigg [ \frac{1}{2 } \left ( \left ( { \mathbf{d}}^\mu { \mathbf{h}}\right ) ^\dagger \otimes { \mathbf{d}}^\nu { \mathbf{h}}+ \left ( { \mathbf{d}}^\nu { \mathbf{h}}\right ) ^\dagger \otimes { \mathbf{d}}^\mu { \mathbf{h}}\right ) - \frac{c_x}{4 } g^{\mu \nu } \left ( { \mathbf{d}}_\rho { \mathbf{h}}\right ) ^\dagger \otimes { \mathbf{d}}^\rho { \mathbf{h}}\notag \\ & & & + \frac{1}{8 } \left ( { \operatorname{tr}\left[\left ( { \mathbf{d}}^\mu { \mathbf{h}}\right ) ^\dagger { \mathbf{d}}^\nu { \mathbf{h}}\right ] } - \frac{c_x}{4 } g^{\mu \nu } { \operatorname{tr}\left[\left ( { \mathbf{d}}_\rho { \mathbf{h}}\right ) ^\dagger { \mathbf{d}}^\rho { \mathbf{h}}\right ] } \right ) \bigg ] \tau^{aa } \ , . \end{aligned}\ ] ]
for a first estimate of the impact of generic resonances to vector - boson scattering processes at the lhc , we study the on - shell nambu - goldstone boson scattering amplitudes . when treating vector - boson scattering as @xmath127 process of massless scalars at high energies , it is convenient to describe kinematic dependencies using mandelstam variables @xmath128 . using custodial symmetry and crossing symmetries , the different @xmath129 nambu - goldstone boson scattering amplitudes
are determined by the master amplitudes @xmath130 . in the gaugeless limit
, the amplitudes for the resonance multiplets @xmath60 , @xmath19 , @xmath119 , and @xmath120 are calculated in the gaugeless limit via the feynman rules given in appendix [ appendix : feynman_rules ] .
[ eq : sigma_amplitudes ] @xmath131 [ eq : phi_amplitudes ] @xmath132 [ eq : tensor_amplitudes ] @xmath133 here and in the following , @xmath134/s^2 $ ] is the second order legendre polynomial in terms of the mandelstam variables .
[ eq : isotensor_tensor_amplitudes_gaugeless ] @xmath135 since the leading - order amplitudes as listed above are unbounded both at the pole and at high energy , we use the t - matrix scheme @xcite to restore unitarity . in order to implement the scheme in @xcite
, we decompose the amplitudes into isospin - spin eigenamplitudes ( the @xmath136-wave , @xmath137-wave and @xmath138-wave kinematic functions @xmath139 , @xmath140 and @xmath141 can be found in appendix [ appendix : partialwaves ] ) : [ eq : isospin - spin - scalar ] @xmath142 [ eq : isospin - spin - isotensor ] @xmath143 [ eq : isospin - spin - tensor ] @xmath144 [ eq : isospin - spin - isotensor - tensor ] @xmath145 as argued below in section [ sec : tensor - unitary - gauge ] , for the numerical off - shell calculation of scattering processes we will need approximate values for the resonance decay widths . if suffices to compute those in the gaugeless limit .
contributions proportional to the masses of the vector bosons and the higgs boson are assumed to be small at high resonance masses and are therefore neglected .
[ eq : gamma_scalar ] @xmath146 for later convenience , we compute the coefficients of the effective dimension - eight operators @xmath147 and @xmath148 @xcite , [ eq : dim8_operator_s ] @xmath149 } { \operatorname{tr}\left [ \left ( { \mathbf{d}}^\mu { \mathbf{h}}\right ) ^\dagger { \mathbf{d}}^\nu { \mathbf{h}}\right ] } , \\ \label{eq : ll - s1 } { \mathcal{l}}_{s,1}&= & & f_{s,1}\ & & { \operatorname{tr}\left [ \left ( { \mathbf{d}}_\mu { \mathbf{h}}\right ) ^\dagger { \mathbf{d}}^\mu { \mathbf{h}}\right ] } { \operatorname{tr}\left [ \left ( { \mathbf{d}}_\nu { \mathbf{h}}\right ) ^\dagger { \mathbf{d}}^\nu { \mathbf{h}}\right]}.\end{aligned}\ ] ] which result from integrating out the resonances @xmath150 , one at a time .
@xmath151 beyond the resonance ,
the nambu - goldstone bosons scattering amplitudes rise proportional to powers of the invariant mass of the scattering system .
they eventually violate unitarity at a certain energy , depending on the resonance coupling .
computing the @xmath152 amplitude in the presence of an isoscalar tensor resonance , for instance , @xmath153 we observe that choosing @xmath154 results in a high degree of divergence .
this is due to contributions of the vector and scalar degree of freedoms in the stckelberg parameterization for the tensor coupled to the derivatives of the current and .
as discussed above , such terms can be written in a non - resonant form and should be interpreted as coefficients of undetermined higher - dimensional local operators . setting thus @xmath155 , we obtain an amplitude @xmath156 which rises proportional to @xmath118 beyond the resonance .
however , the scalar and vector degree of freedoms provide additional contributions which are not manifest in the gaugeless limit .
a calculation of the tensor scattering amplitude in the unitary gauge is necessary .
the longitudinal on - shell @xmath157 amplitude for @xmath158 is given by @xmath159 -1 \right)s^2 + 12 m_w^2 m_z^2 \notag \\ & \phantom{-\frac{1}{24}\frac{f_f^2}{s - m_f^2 } } \quad -12\frac { m_w^2 m_z^2}{m_f^2 } \notag\\ & \phantom{-\frac{1}{24}\frac{f_f^2}{s - m_f^2 } } \notag \quad + \left(s-2 m_w^2\right ) \left(s-2 m_z^2\right ) \\ & \phantom{-\frac{1}{24}\frac{f_f^2}{s - m_f^2 } } \quad \notag + 4\frac { m_w^2 m_z^2}{m_f^4 } s^2 + 2\frac{\left ( m_w^2 + m_z^2 \right)s^2 -4 m_z^2m_w^2 s}{m_f^2 } \bigg ] \ , .\end{aligned}\ ] ] the first line represents the tensor contribution in the stckelberg parameterization . due to its suppression by a power of @xmath118 ,
the vector part in the second line can be neglected for the longitudinal scattering amplitude . besides the scalar contribution originating from the trace of the current , additional contributions related to the double derivative of the current and its mixing with the trace part written in the fourth line will rise with energy .
however , they are suppressed by @xmath160 or @xmath161 and can be neglected if the mass of the tensor resonance is large in comparison to the vector boson masses . in this case
, the longitudinal amplitude of the vector bosons calculated in the unitary gauge coincides with the amplitude in gaugeless limit .
furthermore , due to the coupling to the derivatives of the scalar and vector degrees of freedom , also amplitudes in channels with transverse polarization rise with the energy of the vector - boson scattering system . a full list of these channels in the high - energy limit is displayed in table [ table : tensor - polarization ] .
we observe that all channels which include at least one transversally polarized vector boson are suppressed by @xmath160 .
therefore , a calculation within the gaugeless limit is sufficient to estimate the high - energy behavior for high masses of the tensor resonance .
@xmath162 ( + , & \hspace{0ex}+,&\hspace{0ex}-,-&\hspace{0ex } ) \\[-3 mm ] ( -,&\hspace{0ex}-,&\hspace{0ex}+,+&\hspace{0ex } ) \\[-3 mm ] ( -,&\hspace{0ex}-,&\hspace{0ex}-,- & \hspace{0ex } ) \end{array } & \hspace{1 cm } & & - \frac{c_f^2m_w^2 m_z^2}{24m_f^2 } f_f^2 s \\ \hline \begin{array}{l l l l } ( + , & \hspace{0ex}0,&\hspace{0ex}0,+ & \hspace{0ex } ) \\[-3 mm ] ( 0,&\hspace{0ex}+,&\hspace{0ex}+,0&\hspace{0ex } ) \\[-3 mm ] ( 0,&\hspace{0ex}-,&\hspace{0ex}-,0&\hspace{0ex } ) \\[-3 mm ] ( -,&\hspace{0ex}0,&\hspace{0ex}0,- & \hspace{0ex } ) \end{array } & \hspace{1 cm } & & \frac{m_w m_z}{8m_f^2 } f_f^2 t \\ \hline \begin{array}{l l l l } ( + , & \hspace{0ex}0,&\hspace{0ex}+,0 & \hspace{0ex } ) \\[-3 mm ] ( 0,&\hspace{0ex}+,&\hspace{0ex}0,+&\hspace{0ex } ) \\[-3 mm ] ( 0,&\hspace{0ex}-,&\hspace{0ex}0,-&\hspace{0ex } ) \\[-3 mm ] ( -,&\hspace{0ex}0,&\hspace{0ex}-,0 & \hspace{0ex } ) \end{array } & \hspace{1 cm } & & \frac{m_w m_z}{8m_f^2 } f_f^2 u \\
\hline \begin{array}{l l l l } ( + , & \hspace{0ex}0,&\hspace{0ex}-,0 & \hspace{0ex } ) \\[-3 mm ] ( 0,&\hspace{0ex}+,&\hspace{0ex}0,-&\hspace{0ex } ) \\[-3 mm ] ( 0,&\hspace{0ex}-,&\hspace{0ex}0,+&\hspace{0ex } ) \\[-3 mm ] ( -,&\hspace{0ex}0,&\hspace{0ex}+,0 & \hspace{0ex } ) \end{array } & \hspace{1 cm } & & -\frac{m_w m_z}{8m_f^2 } f_f^2 t \\
\hline \begin{array}{l l l l } ( + , & \hspace{0ex}0,&\hspace{0ex}0,- & \hspace{0ex } ) \\[-3 mm ] ( 0,&\hspace{0ex}+,&\hspace{0ex}-,0&\hspace{0ex } ) \\[-3 mm ] ( 0,&\hspace{0ex}-,&\hspace{0ex}+,0&\hspace{0ex } ) \\[-3 mm ] ( -,&\hspace{0ex}0,&\hspace{0ex}0,+ & \hspace{0ex } ) \end{array } & \hspace{1 cm } & & -\frac{m_w m_z}{8m_f^2 } f_f^2 u \\ \hline \begin{array}{l l l l } ( + , & \hspace{0ex}+,&\hspace{0ex}0,0 & \hspace{0ex } ) \\[-3 mm ] ( -,&\hspace{0ex}-,&\hspace{0ex}0,0&\hspace{0ex } ) \\
\end{array } & \hspace{1 cm } & & \begin{aligned } & \frac { m_f^2 + 2 m_z^2 } { 12 m_f^4 } m_w^2 f_f^2 s \end{aligned } \\
\hline \begin{array}{l l l l } ( 0 , & \hspace{0ex}0,&\hspace{0ex}+,+ & \hspace{0ex } ) \\[-3 mm ] ( 0,&\hspace{0ex}0,&\hspace{0ex}-,-&\hspace{0ex } ) \\
\end{array } & \hspace{1 cm } & & \begin{aligned } & \frac { m_f^2 + 2 m_w^2 } { 12 m_f^4 } m_z^2 f_f^2 s \end{aligned } \\
\hline \begin{array}{l l l l } ( 0 , & \hspace{0ex}0,&\hspace{0ex}0,0 & \hspace{0ex } ) \\ \end{array } & \hspace{1 cm } & & \begin{aligned } \frac{f_f^2}{24 } \frac{2s^2- 3t^2 - 3u^2 } { s } + \frac { m_f^2 \left(m_w^2+m_z^2 \right ) + 2m_w^2m_z^2 } { 12 m_f^4 } f_f^2 s \end{aligned } \\
\hline\end{aligned}\ ] ] for the tensor - isotensor amplitude , the analogous result with @xmath163 is @xmath164 containing t - channel and u - channel contributions , as expected .
the tree - level exchange amplitudes that directly result from evaluating feynman rules , exhibit two distinct sources of unitarity violation .
firstly , the amplitude develops a pole at the resonance mass , on the real axis .
secondly , terms that rise with energy asymptotically violate unitarity bounds . in principle , the t - matrix unitarization scheme would be sufficient to regulate both issues simultaneously . at the pole , this boils down to standard dyson resummation , introducing the particle width as an imaginary part in the denominator .
it can easily be verified that this actually happens for the on - shell scattering amplitudes of external nambu - goldstone bosons .
we obtain the correct value for the resonance width in the gaugeless limit . however , we want to evaluate the amplitudes off - shell for physical @xmath1 and @xmath2 bosons .
the simplified unitarization scheme that we describe above is not exactly accurate as soon as we include finite corrections due to transversal gauge bosons and finite @xmath10 mass . as a result , there are contributions which are not cancelled on the resonance pole , and a narrow but unbounded peak remains . to avoid this problem , we simply insert an _ a priori _ width in the resonant propagator .
we thus start from a _ complex _ model amplitude .
therefore , we take the t - matrix scheme of @xcite at face value , and drop the reference to the usual k - matrix scheme which implies an intermediate projection onto the real axis . by construction , in the gaugeless limit , the correct result is invariant with respect to the introduction of this width , if it has the correct on - shell value . for finite gauge couplings and masses , the result acquires a subleading dependence on this initial value since the model amplitude is neither on the real axis nor exactly on the argand circle .
however , the amplitude after unitarization is now bounded near the resonance pole , as required . in the asymptotic regime
, the simplified t - matrix scheme renders the amplitude unitary at all energies , if the exchanged resonance is scalar .
this enables us to compute cross sections and generate event samples in this model for complete processes at the lhc ( cf .
section [ sec : results ] ) . for a tensor resonance , in the stckelberg approach , the genuine tensor exchange terms
are also regulated completely by this ( simplified ) scheme .
the extra stckelberg vector and scalar terms , however , generate higher powers of @xmath118 which enter when trading nambu - goldstone bosons for physical vector bosons in unitary gauge , suppressed by powers of @xmath165 . applying the unitarization framework for those extra terms would require a complete diagonalization of all vector - boson helicity amplitudes in unitary gauge . in any case , parameter ranges where these terms play a role correspond to a regime where all degrees of freedom of the sm interact strongly via these couplings .
we therefore stay away from this range and choose parameters where those terms are subleading within the accessible energy range .
computing the scale where the stckelberg vector - scalar terms violate the relevant unitarity bounds , we obtain the energy limit @xmath166 for the model which contains an isoscalar tensor , and @xmath167 for the isotensor tensor multiplet . here
, @xmath168 indicates the common mass scale of electroweak bosons @xmath169 . inserting the accessible energy for the lhc collider
, we can invert those relations to extract parameter regions where the simplified models with a tensor resonance are valid .
the numerical results in the following sections have been obtained for parameter values that satisfy the bounds .
in the previous section , we have derived the analytic expressions that determine the on - shell vbs amplitudes in the presence of a resonance .
the amplitudes include correction terms that enforce quantum - mechanical unitarity without altering the physical content of the model . ultimately , we are interested in measurable effects in lhc data . for a complete calculation ,
the unitarized amplitudes that are originally defined for on - shell vbs processes , have to be extrapolated off - shell in a practically meaningful way .
as long as the kinematical conditions are approximately met , we can evaluate the interactions in unitary gauge , eliminating all explicit references to nambu - goldstone bosons in favor of physical vector fields , and derive the feynman rules in that gauge .
the effective feynman rules for the unitarity corrections become momentum dependent and involve theta functions that restrict the insertions to the @xmath118-channel of vbs where partial - wave projection and unitarization is defined . in the physical processes at the lhc ,
@xmath170 where @xmath171 generically denotes a quark and @xmath172 is either @xmath1 or @xmath2 , the final - state quarks are detected as jets in the forward direction . with suitable cuts , we can arrange that there is significant contribution from the subprocess @xmath113 where the initial - state vector bosons are spacelike but approximately on - shell , in the limit of high invariant @xmath173 mass .
this subprocess , i.e. , the associated off - shell amplitude , obtains contributions from resonance exchange and is affected by unitarization .
we have implemented this prescription as a model in the monte - carlo integration and event generation package whizard @xcite .
this is a universal event generator for simulations at hadron and lepton colliders at leading order and next - to - leading ( qcd ) @xcite order .
though interfaces to automated tools for beyond the sm models exist @xcite , they can not be used for the implementation of unitarization projections for operators and resonances .
the reason is the global structure of the unitarization projection .
therefore the models described in the current paper have been manually added to the framework . for each resonance type ( @xmath174 )
, we can compute the relation of the resonance width ( section [ sec : width ] ) to the operator coefficients in the low - energy eft ( section [ sec : matching ] ) which result when the resonance is integrated out .
these relations are listed in table [ tab : gamma ] .
the analysis of lhc run - i data by the atlas experiment @xcite has been cast into bounds on the eft parameters @xmath175 and @xmath176 , namely @xmath177 where only one parameter was varied at a time .
this analysis covered the same - sign leptonic decay channel of @xmath178 and @xmath179 .
it was based on the t - matrix unitarized version of the extrapolated eft as its reference model , with the pure sm as the limit for vanishing parameters .
a cms analysis can be found in @xcite in the following , we will present results both for on - shell @xmath10 final states and for complete partonic final states . on -
shell vector bosons can not be detected directly but their distributions directly reflect the actual features of the physical model .
observable distributions of fermions in the final state , which may be quarks ( jets ) , charged leptons , or neutrinos , are less directly linked to the physical process and require detailed analysis along the lines of @xcite .
this concerns , in particular , the separation of signal and background based on detector data , which is beyond the scope of the present paper .
we show results for particular parameter sets where we add one resonance at a time on top of the sm , namely a scalar - isoscalar , tensor - isoscalar , or scalar - isotensor resonance , respectively .
all extra higher - dimensional operator coefficients are set to zero . by varying the resonance parameters within reasonable limits
, this gives an overview of the expected phenomenology .
for definiteness , we choose to plot the invariant mass of the vector - boson pair system in the final state , which is the energy scale of the actual vbs process .
the initial state is convoluted with the parton structure functions , so the results hold for the lhc ( @xmath180 ) , and we apply standard vbs cuts to enhance the signal .
the final - state vector bosons are taken on - shell .
we show the distribution for the @xmath178 and @xmath181 final states , where the latter case as the _ golden channel _ of vbs is distinguished by the fact that the @xmath181 invariant mass can be reconstructed from the leptonic @xmath2 decays .
this is not possible for @xmath178 , but the corresponding same - sign lepton channel is distinguished by a favorable signal - to - background ratio .
note that in the on - shell plots , the vector - boson decay branching ratios have not been included . in all invariant - mass plots , we display the distribution for the unitarized resonance model ( blue curves ) together with the pure sm prediction ( black ) .
we also plot the unitarity bound for the appropriate partial wave , extrapolated off - shell by the same algorithm , as a dashed curve ( black ) . for illustrative purposes ,
we also display , in each case , the unitarized extrapolation of the low - energy eft ( red , solid ) , where we choose the operator coefficients equal to the formal result of integrating out the resonance .
finally , we also display numerical results for the eft without unitarization ( red , dashed ) and the resonance with correct width but no further unitarization ( blue , dashed ) .
the simplest case is a scalar - isoscalar resonance .
this is a single isolated resonance , as it could arise , e.g. , as the extra scalar particle in a singlet - doublet higgs model or as a low - energy signal of a strongly interacting higgs sector that is neutral under the sm gauge group . and
@xmath182 , for the processes @xmath183 ( left ) and @xmath184 ( right ) , respectively . in the lower plot
, there is a low lying isoscalar scalar with @xmath185 and @xmath186 for the process @xmath184 . solid line : unitarized results , dashed lines : naive result , black dashed line : limit of saturation of @xmath187 @xmath188 or @xmath189 @xmath190 , respectiveluy .
cuts : @xmath191 gev ; @xmath192 ; @xmath193 gev ; @xmath194 .
, title="fig : " ] and @xmath182 , for the processes @xmath183 ( left ) and @xmath184 ( right ) , respectively . in the lower plot
, there is a low lying isoscalar scalar with @xmath185 and @xmath186 for the process @xmath184 . solid line : unitarized results , dashed lines : naive result , black dashed line : limit of saturation of @xmath187 @xmath188 or @xmath189 @xmath190 , respectiveluy .
cuts : @xmath191 gev ; @xmath192 ; @xmath193 gev ; @xmath194 .
, title="fig : " ] + and @xmath182 , for the processes @xmath183 ( left ) and @xmath184 ( right ) , respectively . in the lower plot
, there is a low lying isoscalar scalar with @xmath185 and @xmath186 for the process @xmath184 . solid line : unitarized results , dashed lines : naive result , black dashed line : limit of saturation of @xmath187 @xmath188 or @xmath189 @xmath190 , respectiveluy .
cuts : @xmath191 gev ; @xmath192 ; @xmath193 gev ; @xmath194 .
, title="fig : " ] in fig .
[ fig : isoscalar - scalar ] , upper row , we have selected a moderate mass of @xmath195 and a rather narrow width of @xmath196 , which corresponds to a weak coupling .
the isolated resonance is clearly visible in the @xmath181 channel , while the @xmath178 channel is barely affected .
for such weak coupling , the operator coefficient in the eft is small and more than one order of magnitude below the current lhc run - i limit .
we can draw the conclusion that in this case the resonance should be detectable for sufficient luminosity , but the eft approximation is not useful . turning to a stronger coupling
, we show the corresponding distribution in the @xmath181 channel for @xmath197 and @xmath198 in fig .
[ fig : isoscalar - scalar ] , lower row . here , the eft parameters are within the range that should become accessible at lhc run ii and beyond .
the eft curve ( red , solid ) appears correctly as the taylor expansion of the resonance curve ( blue ) for low energy .
however , the energy region where the deviation from the sm becomes sizable , already coincides with the resonance peak region , so the eft considerably underestimates the event yield . beyond the resonance ,
the eft misses the fact that the distribution falls down again , approaching the sm prediction ( black ) from above .
the result also demonstrates that the additional unitarization of the scalar resonance beyond the breit - wigner approximation with constant width is essential , as is seen by comparing the blue and blue - dashed curves . the naive eft result without unitarization ( red , dashed ) grossly overshoots all conceivable models , which should not cross the unitarity limit ( black - dashed ) . as can be observed from table [ tab : gamma ] , a tensor resonance has a stronger impact on the low - energy eft than a scalar resonance of equal width . in fig .
[ fig : isoscalar - tensor ] , upper row , we display the distributions for a tensor isoscalar resonance with mass @xmath199 and width @xmath200 . and @xmath201 for the processes @xmath183 ( left ) , and @xmath202 ( right ) , respectively .
the lower plot is for a strongly interacting isoscalar tensor with @xmath203 and @xmath204 .
solid line : unitarized results , dashed lines : naive result , black dashed line : limit of saturation of @xmath205 @xmath188 or @xmath206 @xmath190 , respectively .
cuts are the same as in fig .
[ fig : isoscalar - scalar].,title="fig : " ] and @xmath201 for the processes @xmath183 ( left ) , and @xmath202 ( right ) , respectively .
the lower plot is for a strongly interacting isoscalar tensor with @xmath203 and @xmath204 .
solid line : unitarized results , dashed lines : naive result , black dashed line : limit of saturation of @xmath205 @xmath188 or @xmath206 @xmath190 , respectively .
cuts are the same as in fig .
[ fig : isoscalar - scalar].,title="fig : " ] + and @xmath201 for the processes @xmath183 ( left ) , and @xmath202 ( right ) , respectively .
the lower plot is for a strongly interacting isoscalar tensor with @xmath203 and @xmath204 .
solid line : unitarized results , dashed lines : naive result , black dashed line : limit of saturation of @xmath205 @xmath188 or @xmath206 @xmath190 , respectively .
cuts are the same as in fig .
[ fig : isoscalar - scalar].,title="fig : " ] the resonance visibly modifies the distribution already at low energy , such that the eft analysis , given sufficient sensitivity , should catch the deviation from the sm .
nevertheless , the excess at the peak in the @xmath181 channel is sizable . beyond the resonance ,
unitarization is essential in the tensor case . in the @xmath178 final state
the tensor enters only as t - channel exchange , so there is no resonance but a broad enhancement .
this enhancement is rather well described by the corresponding unitarized eft . as in the scalar case
, the curves without unitarization do not provide a useful phenomenological description . in fig .
[ fig : isoscalar - tensor ] , lower row , we consider a heavy tensor - isoscalar with strong coupling , @xmath207 and @xmath208 .
the resonance peak appears as a broad enhancement , which extends to both low and high energies .
the eft approximation , with sizable coefficients , is rather accurate in this case .
the actual resonance curve shows a nontrivial threshold structure which corresponds to the interplay of all partial waves which are excited by s - channel and t - channel exchange contributions .
however , we should keep in mind that the prediction for such a strong coupling is uncertain in any case and should not be taken too seriously . turning to the isotensor case
, we now get a resonance in all final states including @xmath178 .
this is illustrated by the plots in fig .
[ fig : isotensor - scalar ] for @xmath209 and @xmath210 . and @xmath211 for the processes @xmath183 ( left ) , and @xmath202 ( right ) , respectively .
the lower plot shows a low - lying isotensor scalar with @xmath212 and @xmath213 for the process @xmath214 .
solid line : unitarized results , dashed lines : naive result , black dashed line : limit of saturation of @xmath187 @xmath188 or @xmath189 @xmath190 , respectively .
cuts are the same as in fig .
[ fig : isoscalar - scalar].,title="fig : " ] and @xmath211 for the processes @xmath183 ( left ) , and @xmath202 ( right ) , respectively .
the lower plot shows a low - lying isotensor scalar with @xmath212 and @xmath213 for the process @xmath214 .
solid line : unitarized results , dashed lines : naive result , black dashed line : limit of saturation of @xmath187 @xmath188 or @xmath189 @xmath190 , respectively .
cuts are the same as in fig .
[ fig : isoscalar - scalar].,title="fig : " ] + and @xmath211 for the processes @xmath183 ( left ) , and @xmath202 ( right ) , respectively .
the lower plot shows a low - lying isotensor scalar with @xmath212 and @xmath213 for the process @xmath214 .
solid line : unitarized results , dashed lines : naive result , black dashed line : limit of saturation of @xmath187 @xmath188 or @xmath189 @xmath190 , respectively .
cuts are the same as in fig .
[ fig : isoscalar - scalar].,title="fig : " ] due to the large number of degrees of freedom ( nine states which are degenerate in mass ) , the peak is rather prominent while the low - energy eft parameters are again small .
we observe that the peak value is slightly below ( @xmath178 ) and above ( @xmath181 ) the appropriate unitarity limit , respectively .
this is the effect of @xmath215-channel exchange which also contributes and can have either sign .
contrary to the weakly interacting scenario , a non - unitarized low - lying and strongly interacting isotensor - scalar with mass of @xmath216 and width @xmath217 violates the @xmath218 slightly above the resonance as illustrated in fig .
[ fig : isotensor - scalar ] .
therefore , a unitarization is needed for this strongly interacting resonance .
the low - energy effective field theory approach does only coincide in the unitarized case at high energies , because the eigenamplitudes of the isotensor - scalar as well as the dimension - eight operators are already saturated through the t - matrix formalism .
similarly to the isotensor - scalar , every vector - boson scattering channel receives a resonant contribution from the isotensor - tensor multiplet .
the @xmath178 and @xmath181 channel distributions of the isotensor - tensor resonance with mass @xmath219 width @xmath220 are plotted in fig .
[ fig : isotensor - tensor ] , upper row . due to the bound of equation
, the mass of the isotensor - tensor has to be chosen slightly higher than the mass of the isoscalar - tensor in fig .
[ fig : isoscalar - tensor ] when leaving the ratio of width and mass invariant . and @xmath221 for the processes @xmath183 ( left ) , and @xmath202 ( right ) , respectively .
the lower plot shows a strongly interacting isotensor tensor with @xmath222 and @xmath223 for the process @xmath214 .
solid line : unitarized results , dashed lines : naive result , black dashed line : limit of saturation of @xmath205 @xmath188 or @xmath206 @xmath190 , respectively .
cuts are the same as in fig .
[ fig : isoscalar - scalar].,title="fig : " ] and @xmath221 for the processes @xmath183 ( left ) , and @xmath202 ( right ) , respectively .
the lower plot shows a strongly interacting isotensor tensor with @xmath222 and @xmath223 for the process @xmath214 .
solid line : unitarized results , dashed lines : naive result , black dashed line : limit of saturation of @xmath205 @xmath188 or @xmath206 @xmath190 , respectively .
cuts are the same as in fig .
[ fig : isoscalar - scalar].,title="fig : " ] + and @xmath221 for the processes @xmath183 ( left ) , and @xmath202 ( right ) , respectively .
the lower plot shows a strongly interacting isotensor tensor with @xmath222 and @xmath223 for the process @xmath214 .
solid line : unitarized results , dashed lines : naive result , black dashed line : limit of saturation of @xmath205 @xmath188 or @xmath206 @xmath190 , respectively .
cuts are the same as in fig .
[ fig : isoscalar - scalar].,title="fig : " ] the effective field theory with the dimension - eight operators coincides with the onset of the isotensor - tensor peak . starting slightly below the resonance , the resonant cross section deviates from the effective field theory description .
analogously to the isotensor - scalar , the very distinctive peak of the isotensor - tensor is not captured by the dimension - eight operators . in the @xmath178- channel
, even the non - unitarized resonance contribution stays within the unitarity bound of @xmath224 .
contrary to the isotensor - scalar , the isotensor - tensor needs unitarization for the @xmath181 final state due to the large tensor contributions in the @xmath215- and @xmath225channel .
the non - unitarized amplitudes violate the @xmath226 unitarity already below the mass of the resonance .
even the resonance peak is hardly visible .
the unitarized resonance curve shows a peak , although it is slightly above the unitarity bound . in a strongly interacting scenario ( @xmath227 ) , the unitarized isotensor - tensor resonance peaks below its actual mass at @xmath228 .
this peak originates from the already saturated eigenamplitudes , which then fall due to the parton distribution functions at high energies . besides the resonance peak , the low - energy effective field theory coincides with the isotensor - tensor for both unitarized and non - unitarized results .
this is shown in the lower plot of fig .
[ fig : isotensor - tensor ] .
the actual analysis of lhc data will have to exploit cross sections and distributions for the complete final state which consists of the two tagging jets and the decay products of the vector bosons . in this paper
, we only investigate the @xmath181 channel with its decay into four leptons , selecting the @xmath229 final state .
this process is straightforward to analyze , but suffers from the low leptonic branching ratio , so for our simulation we assume the high - luminosity mode of the lhc with integrated luminosity of @xmath230 . we anticipate that by including also the leptonic @xmath231 final state and hadronic final states , the results can be considerably improved .
the simulation generates event samples for the complete process with all feynman graphs , so there is no restriction on resonant vector bosons as the origin of the final - state leptons .
we apply standard vbs cuts and compare , in fig .
[ fig : eemmjj ] , various distributions for the sm ( blue ) , resonance model with a single isoscalar - scalar ( red ) , and the unitarized low - energy eft ( purple ) . [ cols="^,^ " , ] the resonance with mass @xmath232 and width @xmath233 appears , as expected , in the invariant mass distribution and , more indirectly , in other plots . clearly , this parameter set is at the margin of observability in this single channel .
the situation obviously improves if we consider resonances with lower mass , larger coupling , in higher representions , and add other analysis channels .
the higgs sector of the sm , after the discovery of a light higgs , is a new field of study for the experiments at the lhc , and beyond .
while the sm yields precise predictions in accordance with the notion of a weakly coupled theory , a thorough analysis of electroweak data should be guided by reference simplified models which differ from the sm . extending the eft by higher - dimensional operators is useful for analyzing observables with bounded energy , but open scattering data
require enforcing unitarity and extrapolating into a region where perturbation theory in the eft is insufficient . without reference to any particular high - energy model
, we have augmented the eft by resonances with even spin , namely scalar or tensor .
assuming exact @xmath76 gauge invariance and , for simplicity , approximate custodial symmetry both in the eft and beyond , we can distinguish four distinct resonance multiplets with a single free mass and coupling parameter each .
this class of models includes the decoupling limit of multi - higgs models and certain aspects of massive - graviton models .
the models are set up such that we need only take the interaction with the higgs sector into account , while couplings to the gauge and fermion sectors occur only via mixing .
this is consistent with the symmetry assumptions and with our knowledge about electroweak precision data , although it is clearly not guaranteed .
the models allow for arbitrary higher - dimensional operators in the eft , unrelated to resonance exchange , so we do not lose generality .
all amplitude calculations are meaningless unless we enforce quantum - mechanical unitarity , since naive extrapolations yield event rates in the high - energy region that can exceed the unitarity bounds by orders of magnitude .
we have consistently implemented the t - matrix unitarization scheme which works on the complex scattering matrix of the model directly , simplified for the asymptotic range where longitudinal and transveral degrees of freedom decouple .
we have studied the case of a tensor resonance in detail .
since we do not necessarily restrict ourselves to states that are related to gravity , the model differs from the various massive - graviton models and studies that can be found in the literature . to our knowledge , the coupling of a generic tensor resonance to the higgs sector and the resulting predictions for the lhc
have not been considered in detail before .
we find that by employing a stckelberg procedure for the implementation in the lagrangian , instead of the classic fierz - pauli approach , we are able to set up the extended eft for an isolated tensor resonance manifestly separated from non - resonant effects .
scalar and tensor resonances can be handled in close analogy .
it turns out that it is possible to extend an effective theory with an isolated tensor resonance up to a cutoff of order @xmath234 , where @xmath7 is the resonance mass , and @xmath235 is the physical higgs mass . we have implemented the models in the monte - carlo package whizard and computed exemplary distributions and simulated event samples for the lhc .
the numerical results illustrate that resonances in vbs may be detected at the lhc within a certain range of mass and coupling values . for a final verdict
, it will be necessary to perform a complete experimental study and analysis , based on exclusive event samples in combination with background and detector description .
we also find that the comparison with pure - eft results can be misleading if resonance and background can not be clearly separated , as it is typical for the situation at the lhc .
we conclude that data should be analyzed on base of resonance models as well as pure - eft simulations .
this holds , in particular , if limits or values are to be combined between distinct final states or with data obtained at a future lepton collider like the ilc @xcite .
there has been a first study similar to the one presented here , investigating resonances of spins and isospins zero , one and two in 1 tev lepton collisions @xcite , where issues of unitarization did not play a role .
wk and jrr want to thank for the hospitality at the institute for high energy physics ( ihep ) and the center for future high energy physics ( cfhep ) of the chinese academy of sciences at beijing , china , where parts of this work have been completed .
ms acknowledges support by jsps and daad and thanks for the hospitality of the kek theory group during his stay over summer 2015 .
@xmath236 to avoid adding terms proportional to the vacuum expectation value , when adding a higgs pair , we introduce @xmath237 } \rightarrow { \operatorname{tr}\left[\widehat{{\mathbf{h}}^\dagger { \mathbf{h}}}\right]}:= { \operatorname{tr}\left[{\mathbf{h}}^\dagger { \mathbf{h}}- \frac{v^2}{4}\right]}. \label{eq : hhhat}\end{aligned}\ ] ] @xmath238 } & = \left ( \partial^\mu w^\nu_k - \partial^\nu w^\mu_k + g \varepsilon_{ijk } w^\mu_iw^\nu_j \right ) \frac{\tau_k}{2 } \\ & & & = \partial^\mu { \mathbf{w}}^\nu - \partial^\nu { \mathbf{w}}^\mu - { \mathrm{i}}g \left [ { \mathbf{w}}^\mu , { \mathbf{w}}^\nu \right ] , \\ { \mathbf{b}}^{\mu\nu } & \equiv\frac{y}{2 } b^{\mu\nu } & = + \frac{{\mathrm{i}}}{g^\prime } { \left [ d_b^\mu , d_b^\nu \right ] } & = \frac{y}{2 } \left ( \partial^\mu b^\nu - \partial^\nu b^\mu \right ) \\ & & & = \partial^\mu { \mathbf{b}}^\nu - \partial^\nu { \mathbf{b}}^\mu \end{aligned } \label{eq : def - fieldstrengthtensor}\ ] ] the covariant derivative is defined via @xmath239 and @xmath240 the equations of motion for the standard model yield @xmath241}{\mathbf{h}}\ , , \label{eq : eqofmotionh}\\ \left ( { \mathbf{d}}^2 { \mathbf{h}}\right)^\dagger & = \mu^2 { \mathbf{h}}^\dagger - \lambda { \operatorname{tr}\left[{\mathbf{h}}^\dagger { \mathbf{h}}\right]}{\mathbf{h}}^\dagger \ , , \label{eq : eqofmotionhdagger } \\ \partial_\mu { \mathbf{b}}^{\mu\nu } & = -{\mathrm{i}}\frac{g^\prime}{2 } \left ( { \mathbf{h}}^\dagger { \mathbf{d}}^\nu { \mathbf{h}}- \left ( { \mathbf{d}}^\nu { \mathbf{h}}\right ) ^\dagger { \mathbf{h}}\right ) \ , , \\ { \mathbf{d}}_\mu { \mathbf{w}}^{\mu\nu } & = -{\mathrm{i}}\frac{g}{2 } \left ( { \mathbf{d}}^\nu { \mathbf{h}}{\mathbf{h}}^\dagger -{\mathbf{h}}\left({\mathbf{d}}^\nu { \mathbf{h}}\right ) ^\dagger \right ) \end{aligned}\ ] ] the tensor products of pauli matrices for the isospin quintet @xmath242 , the isospin vector @xmath243 , and the isospin scalar @xmath244 are defined , respectively , as [ eq : isospin - eigenvectors ] @xmath245 where the pauli matrix for the isospin singlet is related to @xmath246 all nonzero traces of a product of two tensor products are normalized @xmath247}= { \operatorname{tr}\left[\tau_t^+\tau_t^-\right]}={\operatorname{tr}\left[\tau_t^0\tau_t^0\right]}= { \operatorname{tr}\left[\tau_v^+\tau_v^-\right]}={\operatorname{tr}\left[\tau_v^0\tau_v^0\right]}= { \operatorname{tr}\left[\tau_s\tau_s\right]}=1 \ , .\end{aligned}\ ] ] from the properties of the tensor product @xmath248 and the trace @xmath249 } = { \operatorname{tr}\left[a\right ] } { \operatorname{tr}\left[b\right]}\ ] ] we find @xmath250 } & = { \operatorname{tr}\left[a c\right ] } { \operatorname{tr}\left[b d\right]}\ , .
\end{aligned}\ ] ] this reduces the trace of an isospin singlet @xmath251}=2 { \operatorname{tr}\left[a b\right ] } - { \operatorname{tr}\left[a\right ] } { \operatorname{tr}\left[b\right]}.\end{aligned}\ ] ] multiplying the two pauli matrices related to the isospin singlet leads to @xmath252
the feynman rules which are used to calculate the vector - boson scattering amplitudes are summarized in this appendix . focusing only on weak vector - boson scattering ,
the feynman rules are determined from the lagrangian , where gluons , photons and fermions are omitted .
all lagrangians are defined within the higgs matrix realization whose definition can be found in appendix [ appendix : fields ] .
the standard model lagrangian is given by @xmath253 } -\frac{1}{2}{\operatorname{tr}\left[{\mathbf{b}}_{\mu\nu}{\mathbf{b}}^{\mu\nu}\right ] } \notag \\ & + { \operatorname{tr}\left [ \left ( { \mathbf{d}}_\mu { \mathbf{h}}\right ) ^\dagger { \mathbf{d}}^\mu { \mathbf{h}}\right ] } + \mu^2{\operatorname{tr}\left[{\mathbf{h}}^\dagger { \mathbf{h}}\right ] } -\frac{\lambda}{2}\left ( { \operatorname{tr}\left[{\mathbf{h}}^\dagger { \mathbf{h}}\right ] } \right)^2.\end{aligned}\ ] ] dimension - six and -eight operators affecting only the higgs / nambu - goldstone boson sector are discussed in sections [ sec : matching ] and are given by @xmath254 } \cdot { \operatorname{tr}\left [ \left ( { \mathbf{d}}_\mu { \mathbf{h}}\right ) ^\dagger { \mathbf{d}}^\mu { \mathbf{h}}\right ] } \ , , \\ { \mathcal{l}}_{s,0}&= & & f_{s,0}\ & & { \operatorname{tr}\left [ \left ( { \mathbf{d}}_\mu { \mathbf{h}}\right ) ^\dagger { \mathbf{d}}_\nu { \mathbf{h}}\right ] } \cdot { \operatorname{tr}\left [ \left ( { \mathbf{d}}^\mu { \mathbf{h}}\right ) ^\dagger { \mathbf{d}}^\nu { \mathbf{h}}\right ] } \ , , \label{eq : ll_s0 } \\ \label{eq : ll_s1 } { \mathcal{l}}_{s,1}&= & & f_{s,1}\ & & { \operatorname{tr}\left [ \left ( { \mathbf{d}}_\mu { \mathbf{h}}\right ) ^\dagger { \mathbf{d}}^\mu { \mathbf{h}}\right ] } \cdot { \operatorname{tr}\left [ \left ( { \mathbf{d}}_\nu { \mathbf{h}}\right ) ^\dagger { \mathbf{d}}^\nu { \mathbf{h}}\right ] } \ , .\end{aligned}\ ] ] as an extension to model generic new physics , additional resonances are introduced .
the scalar resonance @xmath60 and the tensor resonance @xmath255 represent singlets of the chiral symmetry group , whereas @xmath256 has the quantum numbers @xmath257 under @xmath258 .
@xmath256 is referred to as isotensor for historical reasons , but it actually includes an isovector @xmath259 and isoscalar @xmath260 besides the isotensor @xmath261 . also the fierz - pauli tensor @xmath119 can be reformulated into a tensor @xmath124 , a vector @xmath123 and a scalar @xmath122 such that canonical propagators can be used for each degree of freedom separately instead of the complicated tensor propagator @xmath262 where the projection operator of spin - two states can be written in terms of the spin - one projection operator , @xmath263 \\ & \qquad\quad - \frac{1}{3 } p^{\mu_1\mu_2}(k , m)p^{\nu_1\nu_2}(k , m ) , \end{aligned}\end{aligned}\ ] ] with @xmath41 the feynman rules in unitary gauge of the lagrangians defined in this paper are listed in this section .
only the relevant vertices for the vector - boson scattering process are shown .
in other words , vertices above four fields for effective operators and above three fields for resonances are neglected .
[ eq : fr_unitary_sm ] @xmath264 \ , , \\
z_{\mu_1}w^+_{\mu_2}w^-_{\mu_3 } : & \hspace{0.5 cm } & -{\mathrm{i}}c_w g & \left [ \left ( p_{1 \ , \mu_3}-p_{2 \ , \mu_3 } \right ) g_{\mu_1\mu_2 } + \left ( p_{3 \ , \mu_2}-p_{1 \ , \mu_2 } \right ) g_{\mu_1\mu_3 } \right.\notag\\ & & & \hspace{2ex}\left .
+ \left ( p_{2 \ , \mu_1}-p_{3 \ , \mu_1 } \right ) g_{\mu_2\mu_3}\right ] \ , , \\
h w^+_{\mu_2}w^-_{\mu_3 } : & \hspace{0.5 cm } & { \mathrm{i}}m_w g&g_{\mu_2 \mu_3 } \
, , \\
h z_{\mu_2}z_{\mu_3 } : & \hspace{0.5 cm } & { \mathrm{i}}m_z g&g_{\mu_2 \mu_3 } \ , , \\
w^+_{\mu_1}w^+_{\mu_2}w^-_{\mu_3}w^-_{\mu_4 } : & \hspace{0.5 cm } & - { \mathrm{i}}g^2 & \left ( g_{\mu_1\mu_4}g_{\mu_2\mu_3 } + g_{\mu_1\mu_3}g_{\mu_2\mu_4}-2 g_{\mu_1\mu_2}g_{\mu_3\mu_4 } \right ) \
, , \\ z_{\mu_1}z_{\mu_2}w^+_{\mu_3}w^-_{\mu_4 } : & \hspace{0.5 cm } & { \mathrm{i}}c_w^2 g^2 & \left ( g_{\mu_1\mu_4}g_{\mu_2\mu_3 } + g_{\mu_1\mu_3}g_{\mu_2\mu_4}-2 g_{\mu_1\mu_2}g_{\mu_3\mu_4 } \right ) \ , , \\
a_{\mu_1}a_{\mu_2}w^+_{\mu_3}w^-_{\mu_4 } : & \hspace{0.5 cm } & { \mathrm{i}}e^2 & \left ( g_{\mu_1\mu_4}g_{\mu_2\mu_3 } + g_{\mu_1\mu_3}g_{\mu_2\mu_4}-2 g_{\mu_1\mu_2}g_{\mu_3\mu_4 } \right ) \ , , \\
a_{\mu_1}z_{\mu_2}w^+_{\mu_3}w^-_{\mu_4 } : & \hspace{0.5 cm } & { \mathrm{i}}e c_wg & \left ( g_{\mu_1\mu_4}g_{\mu_2\mu_3 } + g_{\mu_1\mu_3}g_{\mu_2\mu_4}-2 g_{\mu_1\mu_2}g_{\mu_3\mu_4 } \right ) \ , , \\ h h w^+_{\mu_3}w^-_{\mu_4 } : & \hspace{0.5 cm } & \frac{{\mathrm{i}}}{2 } g^2 & g_{\mu_3\mu_4 } \ , , \\
h h z_{\mu_3}z_{\mu_4 } : & \hspace{0.5 cm } & \frac{{\mathrm{i}}}{2 } \frac{g^2}{c_w^2 } & g_{\mu_3\mu_4}\ , .
\end{aligned}\ ] ] [ [ appendix_rules_unitary_hd ] ] lg ^^ [ eq : fr_unitary_hd ] @xmath265 [ [ appendix_rules_unitary_s ] ] lg ^^ [ eq : fr_unitary_s ] @xmath266 \ , , \\ z_{\mu_1}z_{\mu_2}w^+_{\mu_3}w^-_{\mu_4 } : & \hspace{0.5 cm } & \frac { { \mathrm{i}}g^4 v^4}{16 c_w^2 } & \left [ f_{s,0 } \left ( g_{\mu_1\mu_3}g_{\mu_2\mu_4}+g_{\mu_1\mu_4}g_{\mu_2\mu_3 } \right ) \right . \notag \\ & & & \hspace{2ex}\left . + 2
f_{s,1 } g_{\mu_1\mu_2}g_{\mu_3\mu_4 } \right ] \ , , \\
z_{\mu_1}z_{\mu_2}z_{\mu_3}z_{\mu_4 } : & \hspace{0.5 cm } & \frac { { \mathrm{i}}g^4 v^4}{8 c_w^4 } & \left ( f_{s,0 } + f_{s,1 } \right ) \left [ g_{\mu_1\mu_2}g_{\mu_3\mu_4 } + g_{\mu_1\mu_3}g_{\mu_2\mu_4 } \right .
\notag \\ & & & \hspace{2ex}\phantom{\left ( f_{s,0 } + f_{s,1 } \right ) } \left .
+ g_{\mu_1\mu_4}g_{\mu_2\mu_3 } \right ] \ , , \\
h({p_1})h({p_2})w^+_{\mu_3}w^-_{\mu_4 } : & \hspace{0.5 cm } & -\frac { { \mathrm{i}}g^2 v^2}{4 } & \left [ f_{s,0 } \left ( p_{1 \,\mu_3}p_{2 \,\mu_4}+ p_{1 \,\mu_4}p_{2 \,\mu_3 } \right ) \right .
\notag \\ & & & \hspace{2ex } \left .
+ 2 f_{s,1 } g_{\mu_3\mu_4 } p_1 \cdot p_2 \right ] \ , , \\ h({p_1})h({p_2})z_{\mu_3}z_{\mu_4 } : & \hspace{0.5 cm } & -\frac { { \mathrm{i}}g^2 v^2}{4 c_w^2 } & \left [ f_{s,0 } \left ( p_{1 \,\mu_3}p_{2 \,\mu_4}+ p_{1 \,\mu_4}p_{2 \,\mu_3 } \right ) \right . \notag \\ & & & \hspace{2ex } \left .
+ 2 f_{s,1 } g_{\mu_3\mu_4 } p_1 \cdot p_2 \right ] \ , , \\ h({p_1})h({p_2})h({p_3})h({p_4 } ) : & \hspace{0.5 cm } & 2 { \mathrm{i}}&\left ( f_{s,0 } + f_{s,1 } \right ) \left [ \left ( p_1 \cdot p_2 \right ) \left ( p_3 \cdot p_4 \right ) \right .
\notag \\ & & & \phantom{\left ( f_{s,0 } + f_{s,1 } \right ) } \hspace{2ex } + \left ( p_1 \cdot p_3 \right ) \left ( p_2 \cdot p_4 \right ) \notag \\ & & & \phantom{\left ( f_{s,0 } + f_{s,1 } \right ) } \hspace{2ex } \left .
+ \left ( p_1 \cdot p_4 \right ) \left ( p_2 \cdot p_3 \right ) \right ] \ , .
\end{aligned}\ ] ] [ [ appendix_rules_unitary_sigma ] ] lg ^^ [ eq : fr_unitary_sigma ] @xmath267 [ [ appendix_rules_unitary_phi ] ] lg ^^ [ eq : fr_unitary_phi ] @xmath268 [ [ appendix_rules_unitary_f ] ] lg ^^ [ eq : fr_unitary_f ] @xmath269 \ , , \\ f_{\mu \nu } z_\rho z_\sigma : & \hspace{1 cm } & \frac{{\mathrm{i}}g^2 v^2}{8 c_w^2 } f_f & \left [ g_{\mu \sigma } g_{\nu \rho } + g_{\mu \rho } g_{\nu \sigma } -\frac{c_f}{2 } g_{\mu \nu } g_{\rho \sigma } \right ] \ , , \\ f_{\mu \nu } h \left ( p_1 \right ) h\left ( p_2 \right ) : & \hspace{1 cm } & -\frac{{\mathrm{i}}}{2 } f_f & \left [ p_{1\ , \mu } p_{2\ , \nu } + p_{1\ , \nu } p_{2\ , \mu } -\frac{c_f}{2 } g_{\mu\nu}p_1\cdot p_2 \right ] \ , .\end{aligned}\ ] ] [ [ appendix_rules_unitary_f_stueck ] ] lg ^^ @xmath270 \ , , \\ f_{f\mu \nu } z_\rho z_\sigma : & \hspace{1 cm } & \frac{{\mathrm{i}}g^2 v^2}{8 c_w^2 } f_f & \left [ g_{\mu \sigma } g_{\nu \rho } + g_{\mu \rho } g_{\nu \sigma } -\frac{c_f}{2 } g_{\mu \nu } g_{\rho \sigma } \right ] \ , , \\ f_{f\mu \nu } h \left ( p_1 \right ) h\left ( p_2 \right ) : & \hspace{1 cm } & -\frac{{\mathrm{i}}}{2 } f_f & \left [ p_{1\ , \mu } p_{2\ , \nu } + p_{1\ , \nu } p_{2\ , \mu } -\frac{c_f}{2 } g_{\mu\nu}p_1\cdot p_2 \right]\ , .\end{aligned}\ ] ] because of @xmath271 : @xmath272 \ , .\end{aligned}\ ] ] because of @xmath273 and @xmath274 : @xmath275 \ , , \\ \sigma_f\left ( p \right ) z_\rho z_\sigma : & \hspace{0.5 cm } & \frac{{\mathrm{i}}g^2 v^2}{4\sqrt{6 } c_w^2 } f_f & \left [ \left ( c_f - 1 \right ) g_{\rho \sigma } -\frac{1}{m_f^2}\left ( 2 k_\rho k_\sigma - \frac{c_f}{2 } k^2 g_{\rho\sigma } \right ) \right ] \ , ,
\\ \sigma_f h \left ( p_1 \right ) h\left ( p_2 \right ) : & \hspace{0.5 cm } & -\frac{{\mathrm{i}}}{\sqrt{6 } } f_f & \ , \big [ \left ( c_f - 1 \right ) \left ( p_1 \cdot p_2 \right ) \notag \\ & & & \hspace{2ex } - \frac{1}{m_f^2 } \big ( 2 p_1\cdot\left ( p_1 + p_2 \right ) p_2 \cdot \left ( p_1 + p_2 \right ) \notag\\ & & & \left .
\hspace{2ex } \phantom{- \frac{1}{m_f^2 } \big ( } \hspace{2ex } - \frac{c_f}{2 } p_1\cdot p_2 \left ( p_1 + p_2 \right ) ^2 \big ) \right ] \ , .\end{aligned}\ ] ] [ [ appendix_rules_unitary_x ] ] lg ^^ [ eq : fr_unitary_x ] @xmath276 \ , , \\ x_{t\mu\nu}^{\pm } w^{\mp}_\rho z_\sigma : & \hspace{1 cm } & \frac{{\mathrm{i}}g^2 v^2}{8 \sqrt{2}c_w } f_x & \left [ g_{\mu \sigma } g_{\nu \rho } + g_{\mu \rho } g_{\nu \sigma } -\frac{c_x}{2 } g_{\mu \nu } g_{\rho \sigma } \right ] \ , , \\ x_{t\mu\nu}^{0 } w^{\mp}_\rho w^{\pm}_\sigma : & \hspace{1 cm } & -\frac{{\mathrm{i}}g^2 v^2}{8 \sqrt{6 } } f_x & \left [ g_{\mu \sigma } g_{\nu \rho } + g_{\mu \rho } g_{\nu \sigma } -\frac{c_x}{2 } g_{\mu \nu } g_{\rho \sigma } \right ] \ , , \\ x_{t\mu\nu}^{0 } z_\rho z_\sigma : & \hspace{1 cm } & \frac{{\mathrm{i}}g^2 v^2}{4 \sqrt{6}c_w^2 } f_x & \left [ g_{\mu \sigma } g_{\nu \rho } + g_{\mu \rho } g_{\nu \sigma } -\frac{c_x}{2 } g_{\mu \nu } g_{\rho \sigma } \right ] \ , , \\ x_{s\mu\nu } w^{\mp}_\rho w^{\pm}_\sigma : & \hspace{1 cm } & \frac{{\mathrm{i}}g^2 v^2}{16 \sqrt{3 } } f_x & \left [ g_{\mu \sigma } g_{\nu \rho } + g_{\mu \rho } g_{\nu \sigma } -\frac{c_x}{2 } g_{\mu \nu } g_{\rho \sigma } \right ] \ , , \\ x_{s\mu\nu } z_\rho z_\sigma : & \hspace{1 cm } & \frac{{\mathrm{i}}g^2 v^2}{16 \sqrt{3}c_w^2 } f_x & \left [ g_{\mu \sigma } g_{\nu \rho } + g_{\mu \rho } g_{\nu \sigma } -\frac{c_x}{2 } g_{\mu \nu } g_{\rho \sigma } \right ] \ , , \\ x_{v\mu\nu}^{\pm } h \left(p
\right ) w^\mp_{\rho } : & \hspace{1 cm } & -\frac { g v}{4 \sqrt{2 } } f_x & \left [ { p}_{\mu } g_{\nu\rho } + { p}_{\nu } g_{\mu\rho } - \frac{c_x}{2}{p}_{\rho } g_{\mu\nu } \right]\ , , \\
x_{v\mu\nu } h \left(p
\right ) z_{\rho } : & \hspace{1 cm } & \frac { g v}{4 \sqrt{2}c_w } f_x & \left [ { p}_{\mu } g_{\nu\rho } + { p}_{\nu } g_{\mu\rho } - \frac{c_x}{2}{p}_{\rho } g_{\mu\nu } \right]\ , , \\
x_{s\mu\nu } h \left ( p_1 \right ) h \left ( p_2 \right ): & \hspace{1 cm } & \frac{\sqrt{3 } } { 4 } { \mathrm{i}}f_x & \left [ p_{1\ , \mu } p_{2\ , \nu } + p_{1\ , \nu } p_{2\ , \mu } -\frac{c_x}{2 } g_{\mu\nu}p_1\cdot p_2 \right ] \ , .
\end{aligned}\ ] ] in this appendix we collect expressions appearing in the partial - wave expansion of amplitudes .
[ eq : partialwaves_abbr ] @xmath277
in the t - matrix unitarization scheme , the unitarization corrections are expressed as momentum - dependent counterterms for the use as effective feynman rules in the complete amplitude evaluation .
starting from the spin - isospin eigenamplitudes in the gaugeless limit , section [ sec : ilinwhizard ] , the straightforward application of the algorithm in @xcite yields @xmath118-dependent amplitude corrections @xmath278 .
the insertion as effective feynman rules proceeds in form of the following expressions : [ eq : ct_feynmanrules ] @xmath279 , \\
w^\pm_{\mu_1}w^\mp_{\mu_2}\rightarrow z_{\mu_3}z_{\mu_4 } & : & & \frac{g^4 v^4}{4 c_w^2 } & & \left [ \left ( \frac{1}{3 } \left ( \delta { \mathcal{a}}_{00}(s)-\delta { \mathcal{a}}_{20}(s ) \right ) \right .
\notag \\ & & & & & \left .
\left . -\frac{10}{3}\left
( \delta { \mathcal{a}}_{02}(s)-\delta { \mathcal{a}}_{22}(s ) \right ) \right ) \frac{g_{\mu_1\mu_2}g_{\mu_3\mu_4}}{s^2 } \right . \notag\\ & & & & & \left .
+ 5 \left ( \delta { \mathcal{a}}_{02}(s)-\delta { \mathcal{a}}_{22}(s ) \right ) \frac{g_{\mu_1\mu_3}g_{\mu_2\mu_4}+ g_{\mu_1\mu_4}g_{\mu_2\mu_3}}{s^2 } \right ] , \\ w^\pm_{\mu_1}z_{\mu_2}\rightarrow w^\pm_{\mu_3}z_{\mu_4 } & : & & \frac{g^4 v^4}{4c_w^2 } & & \left [ \left ( \frac{1}{2}\delta { \mathcal{a}}_{20}(s)-5\delta { \mathcal{a}}_{22}(s ) \right ) \frac{g_{\mu_1\mu_2}g_{\mu_3\mu_4}}{s^2 } \right .
\notag\\ & & & & & + \left ( -\frac{3}{2}\delta { \mathcal{a}}_{11}(s)+\frac{15}{2}\delta { \mathcal{a}}_{22}(s ) \right ) \frac{g_{\mu_1\mu_3}g_{\mu_2\mu_4}}{s^2 } \notag\\ & & & & & \left .+ \left ( \frac{3}{2}\delta { \mathcal{a}}_{11}(s)+\frac{15}{2}\delta { \mathcal{a}}_{22}(s ) \right ) \frac{g_{\mu_1\mu_4}g_{\mu_2\mu_3}}{s^2 } \right ] , \\ w^\pm_{\mu_1}w^\mp_{\mu_2}\rightarrow w^\pm_{\mu_3}w^\mp_{\mu_4 } & : & & \frac{g^4 v^4}{4 } & & \left [ \left ( \frac{1}{6 } \left ( 2\delta { \mathcal{a}}_{00}(s)+\delta { \mathcal{a}}_{20}(s )
\right ) \right .
\notag\\ & & & & & \left .
-\frac{5}{3}\left ( 2 \delta { \mathcal{a}}_{02}(s)+\delta { \mathcal{a}}_{22}(s )
\right ) \right ) \frac{g_{\mu_1\mu_2}g_{\mu_3\mu_4}}{s^2 } \right . \notag\\ & & & & & + \left ( 5\delta { \mathcal{a}}_{02}(s)-\frac{3}{2}\delta { \mathcal{a}}_{11}(s)+\frac{5}{2 } \delta { \mathcal{a}}_{22}(s ) \right ) \frac{g_{\mu_1\mu_3}g_{\mu_2\mu_4}}{s^2 } \notag\\ & & & & & \left .
+ \left ( 5\delta { \mathcal{a}}_{02}(s)+\frac{3}{2}\delta { \mathcal{a}}_{11}(s)+\frac{5}{2 } \delta { \mathcal{a}}_{22}(s ) \right ) \frac{g_{\mu_1\mu_4}g_{\mu_2\mu_3}}{s^2 } \right ] , \\ z_{\mu_1}z_{\mu_2}\rightarrow z_{\mu_3}z_{\mu_4 } & : & & \frac{g^4 v^4}{4c_w^4 } & & \left [ \left ( \frac{1}{3 } \left ( \delta { \mathcal{a}}_{00}(s)+2\delta { \mathcal{a}}_{20}(s ) \right ) \right .
\notag\\ & & & & & \left .
\left .
-\frac{10}{3}\left ( \delta { \mathcal{a}}_{02}(s)+2\delta { \mathcal{a}}_{22}(s )
\right ) \right ) \frac{g_{\mu_1\mu_2}g_{\mu_3\mu_4}}{s^2 } \right . \notag\\ & & & & & \left .+ 5 \left ( \delta { \mathcal{a}}_{02}(s)+2\delta { \mathcal{a}}_{22}(s ) \right ) \frac{g_{\mu_1\mu_3}g_{\mu_2\mu_4}+ g_{\mu_1\mu_4}g_{\mu_2\mu_3}}{s^2 } \right ] .\end{aligned}\ ] ] these relations are the generalizations of the corresponding formulae in reference @xcite for the case of resonances .
scattering processes involving a higgs boson have a different off - shell extrapolation .
therefore , the higgs momentum is included in the feynman rules for the analogous effective vertices given by @xmath280 , \\ z^{\mu_1}z^{\mu_2}\rightarrow h h & : & & \;-\frac{g^2 v^2}{c_w^2 } & & \left [ \left ( \frac{1}{3 } \left ( \delta { \mathcal{a}}_{00}(s)-\delta { \mathcal{a}}_{20}(s ) \right ) \right .
\notag \\ & & & & & \left .
\left . -\frac{10}{3}\left
( \delta { \mathcal{a}}_{02}(s)-\delta { \mathcal{a}}_{22}(s ) \right ) \right ) \frac{g^{\mu_1\mu_2 } \left(k_3\cdot k_4\right)}{s^2 } \right . \notag\\ & & & & & \left .
+ 5 \left ( \delta { \mathcal{a}}_{02}(s)-\delta { \mathcal{a}}_{22}(s ) \right ) \frac{k_3^{\mu_1}k_4^{\mu_2}+ k_4^{\mu_1}k_3^{\mu_2}}{s^2 } \right ] , \\ { w^\pm}^{\mu_1}h\rightarrow { w^\pm}^{\mu_3}h & : & & \;-g^2 v^2 & & \left [ \left ( \frac{1}{2}\delta { \mathcal{a}}_{20}(s)-5\delta { \mathcal{a}}_{22}(s ) \right ) \frac{k_2^{\mu_1}k_4^{\mu_3}}{s^2 } \right .
\notag\\ & & & & & + \left ( -\frac{3}{2}\delta { \mathcal{a}}_{11}(s)+\frac{15}{2}\delta { \mathcal{a}}_{22}(s ) \right ) \frac{g^{\mu_1\mu_3}\left ( k_2\cdot k_4 \right)}{s^2 } \notag\\ & & & & & \left .+ \left ( \frac{3}{2}\delta { \mathcal{a}}_{11}(s)+\frac{15}{2}\delta { \mathcal{a}}_{22}(s ) \right ) \frac{k_4^{\mu_1}k_2^{\mu_3}}{s^2 } \right ] , \\ { z}^{\mu_1}h\rightarrow { z}^{\mu_3}h & : & & \;-\frac{g^2 v^2}{c_w^2 } & & \left [ \left ( \frac{1}{2}\delta { \mathcal{a}}_{20}(s)-5\delta { \mathcal{a}}_{22}(s ) \right ) \frac{k_2^{\mu_1}k_4^{\mu_3}}{s^2 } \right .
\notag\\ & & & & & + \left ( -\frac{3}{2}\delta { \mathcal{a}}_{11}(s)+\frac{15}{2}\delta { \mathcal{a}}_{22}(s ) \right ) \frac{g^{\mu_1\mu_3}\left ( k_2\cdot k_4 \right)}{s^2 } \notag\\ & & & & & \left .+ \left ( \frac{3}{2}\delta { \mathcal{a}}_{11}(s)+\frac{15}{2}\delta { \mathcal{a}}_{22}(s ) \right ) \frac{k_4^{\mu_1}k_2^{\mu_3}}{s^2 } \right ] , \\ hh\rightarrow hh & : & & & 4&\left [ \left ( \frac{1}{3 } \left ( \delta { \mathcal{a}}_{00}(s)+2\delta { \mathcal{a}}_{20}(s ) \right ) \right . \right .
\notag\\ & & & & & \left .
\left .
-\frac{10}{3}\left ( \delta { \mathcal{a}}_{02}(s)+2\delta { \mathcal{a}}_{22}(s ) \right ) \right ) \frac{\left ( k_1 \cdot k_2\right ) \left ( k_3 \cdot k_4 \right ) } { s^2 } \right . \\
& & & & & \left .+ 5 \left ( \delta { \mathcal{a}}_{02}(s)+2\delta { \mathcal{a}}_{22}(s )
\right ) \frac{\left ( k_1 \cdot k_4\right ) \left ( k_2 \cdot k_3 \right ) + \left ( k_1 \cdot k_4\right ) \left ( k_2 \cdot k_3 \right ) } { s^2 } \right ] \notag.\end{aligned}\ ] ] 99 t. appelquist and j. carazzone , phys .
d * 11 * ( 1975 ) 2856 .
k. hagiwara , s. ishihara , r. szalapski and d. zeppenfeld , phys .
d * 48 * , 2182 ( 1993 ) .
j. f. gunion and h. e. haber , phys .
d * 67 * ( 2003 ) 075019 [ hep - ph/0207010 ] .
h. e. haber and y. nir , nucl .
b * 335 * ( 1990 ) 363 .
h. e. haber , arxiv:1401.0152 [ hep - ph ] .
y. nambu , phys .
lett . * 4 * , 380 ( 1960 ) .
j. goldstone , nuovo cim . * 19 * , 154 ( 1961 ) .
j. goldstone , a. salam and s. weinberg , phys .
127 * , 965 ( 1962 ) . c. e. vayonakis , lett .
nuovo cim .
* 17 * , 383 ( 1976 ) .
m. s. chanowitz and m. k. gaillard , nucl .
b * 261 * , 379 ( 1985 ) .
g. j. gounaris , r. kogerler and h. neufeld , phys .
d * 34 * , 3257 ( 1986 ) .
y. p. yao and c. p. yuan , phys .
d * 38 * , 2237 ( 1988 ) .
j. bagger and c. schmidt , phys .
d * 41 * , 264 ( 1990 ) .
h. j. he , y. p.
kuang and x. y. li , phys .
lett . *
69 * , 2619 ( 1992 ) .
h. j. he , y. p.
kuang and x. y. li , phys .
b * 329 * , 278 ( 1994 ) [ hep - ph/9403283 ] .
h. j. he , y. p.
kuang and x. y. li , phys .
d * 49 * , 4842 ( 1994 ) .
h. j. he and w. b. kilgore , phys .
d * 55 * , 1515 ( 1997 ) [ hep - ph/9609326 ] .
g. aad et al .
[ atlas collaboration ] , phys .
lett . * 113 * , no .
14 , 141803 ( 2014 ) [ arxiv:1405.6241 [ hep - ex ] ] .
the atlas collaboration [ atlas collaboration ] , atlas - conf-2014 - 013 .
cms collaboration [ cms collaboration ] , cms - pas - fsq-13 - 008 .
s. weinberg , phys .
* 17 * , 616 ( 1966 ) .
m. s. chanowitz , m. golden and h. georgi , phys .
d * 36 * , 1490 ( 1987 ) .
b. w. lee , c. quigg and h. b. thacker , phys .
* 38 * , 883 ( 1977 ) ; phys .
d * 16 * , 1519 ( 1977 ) .
d. espriu and f. mescia , phys .
d * 90 * , no . 1 , 015035 ( 2014 ) [ arxiv:1403.7386 [ hep - ph ] ] .
r. l. delgado , a. dobado and f. j. llanes - estrada , phys .
d * 91 * , no .
7 , 075017 ( 2015 ) [ arxiv:1502.04841 [ hep - ph ] ] .
t. appelquist and c. w. bernard , phys .
d * 22 * , 200 ( 1980 ) .
a. c. longhitano , phys .
d * 22 * , 1166 ( 1980 ) .
d. a. dicus , j. f. gunion and r. vega , phys .
b * 258 * , 475 ( 1991 ) .
v. d. barger , k. m. cheung , t. han and r. j. n. phillips , phys .
d * 42 * , 3052 ( 1990 ) .
m. s. berger and m. s. chanowitz , phys .
b * 263 * , 509 ( 1991 ) .
s. n. gupta , j. m. johnson and w. w. repko , phys .
d * 48 * , 2083 ( 1993 ) [ hep - ph/9307239 ] .
t. appelquist and g. h. wu , phys . rev .
d * 48 * , 3235 ( 1993 ) [ hep - ph/9304240 ] .
m. s. chanowitz , phys .
b * 373 * , 141 ( 1996 ) [ hep - ph/9512358 ] .
j. bagger , v. d. barger , k. m. cheung , j. f. gunion , t. han , g. a. ladinsky , r. rosenfeld and c .- p .
yuan , phys .
d * 52 * , 3878 ( 1995 ) [ hep - ph/9504426 ] .
j. m. butterworth , b. e. cox and j. r. forshaw , phys .
d * 65 * , 096014 ( 2002 ) [ hep - ph/0201098 ] .
o. j. p. eboli , m. c. gonzalez - garcia and s. m. lietti , phys .
d * 69 * , 095005 ( 2004 ) [ hep - ph/0310141 ] .
o. j. p. eboli , m. c. gonzalez - garcia and j. k. mizukoshi , phys . rev .
d * 74 * , 073005 ( 2006 ) [ hep - ph/0606118 ] .
j. distler , b. grinstein , r. a. porto and i. z. rothstein , phys .
lett . * 98 * , 041601 ( 2007 ) [ hep - ph/0604255 ] .
t. han , d. krohn , l. t. wang and w. zhu , jhep * 1003 * , 082 ( 2010 ) [ arxiv:0911.3656 [ hep - ph ] ] .
a. freitas and j. s. gainer , phys .
d * 88 * , no . 1 , 017302 ( 2013 ) [ arxiv:1212.3598 ] . d. espriu and b. yencho , phys .
d * 87 * , no .
5 , 055017 ( 2013 ) [ arxiv:1212.4158 [ hep - ph ] ]
. k. doroba , j. kalinowski , j. kuczmarski , s. pokorski , j. rosiek , m. szleper and s. tkaczyk , phys .
d * 86 * , 036011 ( 2012 ) [ arxiv:1201.2768 [ hep - ph ] ]
. j. chang , k. cheung , c. t. lu and t. c. yuan , phys .
d * 87 * , 093005 ( 2013 ) [ arxiv:1303.6335 [ hep - ph ] ] .
r. l. delgado , a. dobado and f. j. llanes - estrada , arxiv:1412.3277 [ hep - ph ] .
m. fabbrichesi , m. pinamonti , a. tonero and a. urbano , arxiv:1509.06378 [ hep - ph ] .
m. szleper , arxiv:1412.8367 [ hep - ph ] .
a. alboteanu , w. kilian and j. reuter , jhep * 0811 * , 010 ( 2008 ) [ arxiv:0806.4145 [ hep - ph ] ] .
w. kilian , t. ohl , j. reuter and m. sekulla , phys .
d * 91 * ( 2015 ) 096007 [ arxiv:1408.6207 [ hep - ph ] ] .
i. antoniadis , phys .
b * 246 * ( 1990 ) 377 .
n. arkani - hamed , s. dimopoulos and g. r. dvali , phys .
b * 429 * ( 1998 ) 263 [ hep - ph/9803315 ] .
i. antoniadis , n. arkani - hamed , s. dimopoulos and g. r. dvali , phys .
b * 436 * ( 1998 ) 257 [ hep - ph/9804398 ] .
@xcite t. han , j. d. lykken and r. j. zhang , phys .
d * 59 * , 105006 ( 1999 ) [ hep - ph/9811350 ] .
n. bouatta , g. compere and a. sagnotti , hep - th/0409068 .
l. p. s. singh and c. r. hagen , phys .
d * 9 * ( 1974 ) 898 .
i. l. buchbinder and v. a. krykhtin , nucl .
b * 727 * ( 2005 ) 537 [ hep - th/0505092 ] .
s. -z .
huang , p. -f .
zhang , t. -n .
ruan , y. -c .
zhu and z. -p .
zheng , eur .
j. c * 42 * ( 2005 ) 375 .
m. sekulla ph.d .
thesis , university of siegen , 2015 s. weinberg , phys .
* 133 * ( 1964 ) b1318 .
m. fierz and w. pauli , proc .
a * 173 * ( 1939 ) 211 .
n. arkani - hamed , h. georgi and m. d. schwartz , annals phys .
* 305 * ( 2003 ) 96 [ hep - th/0210184 ] .
n. arkani - hamed and m. d. schwartz , phys .
d * 69 * ( 2004 ) 104001 [ hep - th/0302110 ] .
m. d. schwartz , phys .
d * 68 * ( 2003 ) 024029 [ hep - th/0303114 ] . c. de rham , g. gabadadze and a. j. tolley , phys .
b * 711 * ( 2012 ) 190 [ arxiv:1107.3820 [ hep - th ] ] .
s. f. hassan , a. schmidt - may and m. von strauss , phys .
b * 715 * ( 2012 ) 335 [ arxiv:1203.5283 [ hep - th ] ] .
j. noller , j. h. c. scargill and p. g. ferreira , jcap * 1402 * ( 2014 ) 007 [ arxiv:1311.7009 [ hep - th ] ] .
k. hinterbichler , rev .
* 84 * ( 2012 ) 671 [ arxiv:1105.3735 [ hep - th ] ] .
e. c. g. stueckelberg , helv .
acta * 11 * , 225 ( 1938 ) .
e. c. g. stueckelberg , helv .
acta * 15 * ( 1942 ) 23 .
h. ruegg and m. ruiz - altaba , int .
j. mod .
a * 19 * ( 2004 ) 3265 [ hep - th/0304245 ] .
j. bonifacio , p. g. ferreira and k. hinterbichler , phys .
d * 91 * ( 2015 ) 125008 [ arxiv:1501.03159 [ hep - th ] ] .
c. p. burgess and d. london , hep - ph/9203215 .
s. weinberg , phys .
rev . * 166 * , 1568 ( 1968 ) ; w. kilian , springer tracts mod .
phys . * 198 * , 1 ( 2003 ) .
d. b. kaplan and h. georgi , phys .
b * 136 * , 183 ( 1984 ) . for reviews of little higgs models
cf . : m. schmaltz and d. tucker - smith , _ little higgs review _ , ann .
nucl . part .
sci . * 55 * , 229 ( 2005 ) [ hep - ph/0502182 ] ; m. perelstein , _ little higgs models and their phenomenology _ , prog .
phys . * 58 * , 247 ( 2007 ) [ hep - ph/0512128 ] ; ewpt , fine - tuning have been studied here : w. kilian and j. reuter , phys .
d * 70 * , 015004 ( 2004 ) [ hep - ph/0311095 ] .
j. reuter and m. tonini , jhep * 1302 * , 077 ( 2013 ) [ arxiv:1212.5930 [ hep - ph ] ] .
j. reuter , m. tonini and m. de vries , arxiv:1307.5010 [ hep - ph ] ; jhep * 1402 * , 053 ( 2014 ) [ arxiv:1310.2918 [ hep - ph ] ] . w. kilian , t. ohl and j. reuter , _ whizard : simulating multi - particle processes at lhc and ilc _ , eur . phys .
j. c * 71 * , 1742 ( 2011 ) [ arxiv:0708.4233 [ hep - ph ] ] .
m. moretti , t. ohl and j. reuter , _
omega : an optimizing matrix element generator _ , in * 2nd ecfa / desy study 1998 - 2001 * 1981 - 2009 [ hep - ph/0102195 ] .
w. kilian , t. ohl , j. reuter and c. speckner , jhep * 1210 * , 022 ( 2012 ) [ arxiv:1206.3700 [ hep - ph ] ] .
w. kilian , j. reuter , s. schmidt and d. wiesler , jhep * 1204 * , 013 ( 2012 ) [ arxiv:1112.1039 [ hep - ph ] ] .
n. d. christensen , c. duhr , b. fuks , j. reuter and c. speckner , eur .
j. c * 72 * , 1990 ( 2012 ) [ arxiv:1010.3251 [ hep - ph ] ] . b. chokoufe nejad , w. kilian , j. reuter and c. weiss , pos eps * -hep2015 * , 317 ( 2015 ) [ arxiv:1510.02739 [ hep - ph ] ] ; c. weiss , b. chokoufe nejad , w. kilian and j. reuter , pos eps * -hep2015 * , 466 ( 2015 ) [ arxiv:1510.02666 [ hep - ph ] ] ; n. greiner , a. guffanti , t. reiter and j. reuter , phys .
lett . * 107 * , 102002 ( 2011 ) [ arxiv:1105.3624 [ hep - ph ] ] ; t. binoth , n. greiner , a. guffanti , j. reuter , j .-
guillet and t. reiter , phys .
b * 685 * , 293 ( 2010 ) [ arxiv:0910.4379 [ hep - ph ] ] ; w. kilian , j. reuter and t. robens , eur .
j. c * 48 * , 389 ( 2006 ) [ hep - ph/0607127 ] .
s. fichet and g. von gersdorff , arxiv:1508.04814 [ hep - ph ] .
h. baer et al . , arxiv:1306.6352 [ hep - ph ] . t. behnke et al . , arxiv:1306.6329 [ physics.ins-det ] . m. beyer , w. kilian , p. krstonosic , k. mnig , j. reuter , e. schmidt and h. schrder , eur .
j. c * 48 * , 353 ( 2006 ) [ hep - ph/0604048 ] .
|
we study in a bottom - up approach the theoretically consistent description of additional resonances in the electroweak sector beyond the discovered higgs boson as simplified models .
we focus on scalar and tensor resonances .
our formalism is suited for strongly coupled models , but can also be applied to weakly interacting theories .
the spurious degrees of freedom of tensor resonances that would lead to bad high - energy behavior are treated using a generalization of the stckelberg formalism .
we calculate scattering amplitudes for vector - boson and higgs boson pairs .
the high - energy region is regulated by the t - matrix unitarization procedure , leading to amplitudes that are well behaved on the whole phase space .
we present numerical results for complete partonic processes that involve resonant vector - boson scattering , for the current and upcoming runs of lhc .
|
spin qubits are promising candidates for quantum information processing due to their long decoherence and relaxation time.@xcite some schemes , such as solid state spin qbuits , further enjoy the potential scalability via the integration with nanotechnology.@xcite however , spin qubits are not totally immune from the ubiquitous decoherence . to describe the bath that causes the decoherence of spin qubits , it is known that in some cases , the bath is better modeled by spins instead of delocalized oscillators , resulting in the so - called spin baths.@xcite
it has been argued that the influence of spin baths may be qualitatively different from bosonic baths and non - markovian dynamics can easily emerge.@xcite due to the growing interest in spin baths , the decoherence behavior and the entanglement dynamics of few qubits coupled to spin baths have been studied extensively in recent years .
early works focus on the decoherence due to independent spins.@xcite here although the proposed model formally resembles a spin boson model , non - markovian already emerges even when bath modes are not interacting.@xcite in real baths , however , spins are not independent .
it is therefore important to include effects due to interactions of spins in the bath .
nonetheless , the inclusion of the intra - spin interaction in the bath complicates the problem and only for some limited models with high symmetry , exact reduced dynamics can be identified .
@xcite beyond models with exact solutions , approximated dynamics was obtained by using mean - field@xcite or perturbative approaches@xciteto handle more generic models .
the most common model employed in these works is the `` central spin model '' , where the qubits are uniformly coupled to all spins of the bath . while analytical derivations are possible in these models , they are less realistic and are more difficult to be implemented experimentally . a non - perturbative approach that can capture the non - markovian effects induced by interacting spin bath with generic coupling to qubits is hence highly desirable . to overcome the difficulty associated with interacting spins , we utilize the method of time - dependent density matrix renormalization group ( t - dmrg ) @xcite to investigate the reduced dynamics of single or two qubits coupled to an interacting spin chain .
recently , t - dmrg has been used to study the single qubit pure dephasing induced by a @xmath0 anisotropic spin chain.@xcite the advantage of t - dmrg is its ability to calculate reduced dynamics even when the spin bath is not integrable and the coupling is not uniform . due to the accumulation of errors , t - dmrg will eventually run away at large time@xcite but this does not impose serious limitation since for the study of quantum information we are mostly interested in some smaller time scale . in this work
we apply the method of t - dmrg to investigate both pure dephasing and general decoherence model of qubits coupled to spin baths .
single qubit decoherence as well as two qubits ( dis)-entanglement dynamics are investigated .
it is shown that both decoherence and entanglement strongly depend on the phase of the spin bath .
in general , we find that spin baths can induce entanglement for an initially disentangled pair of qubits .
however , when the spin bath is in the ferromagnetic phase , because qubits directly couple to the order parameter , the reduced dynamics shows oscillatory type behavior . on the other hand , only for paramagnetic and antiferromagnetic phases , initially entangled states suffer from the entanglement sudden death.@xcite to quantify the single spin decoherence , we evaluate the evolution of the loschmidt echo.@xcite we analyze the relation between the short time loschmidt echo decay parameter and the quantum phases of the spin bath , as it has been pointed out that these two are closely related , especially when a symmetry breaking occurs in the bath.@xcite we use the temporal evolution of concurrence@xcite to study the entanglement dynamics .
one important issue of entanglement dynamics is the possibility of creating entanglement through a common bath for originally dis - entangled qubits .
it has been shown that induced entanglement via a common bath is possible for bosonic and fermionic baths.@xcite for spin baths such a possibility has been explored for non - interacting spin bath@xcite as well as interacting ones@xcite but is restricted to uniform coupling models . it will be shown later in this paper that induced entanglement is possible for local coupling model considered in this work .
we note that the induced entanglement is also closely related to recent proposals of quantum communication and teleportation via spin chain.@xcite another important issue is the disentanglement dynamics of an initially entangled state .
it has attracted much attention in recent years since yu and eberly@xcite and jakbczyk and jamrz@xcite predicted that two initially entangled state without interaction can become completely disentangled at finite time .
this feature has been termed entanglement sudden death ( esd ) .
esd has been studied theoretically within various models @xcite and has been demonstrate experimentally.@xcite these models , however , are restricted to markovian bosonic bath or classical noise . in this work we explore for the first time if esd - like phenomena can occur for spin baths in non - markovian regime .
in particular , we will show that only when the spin bath are paramagnetic or antiferromagnetic , the phenomenon of esd occurs , while when the spin bath is in the ferromagnetic phase , concurrence shows oscillatory behavior .
as understanding the nature of the ( dis)-entanglement dynamics constitutes an important step for quantum engineering these systems , our results are of practical usage for future quantum information processing .
the paper is organized as follows . in sec .
[ sec : model ] we present our model hamiltonian and briefly discuss how to apply t - dmrg to analyze the model hamiltonian . in sec .
[ sec : single ] we present our results of single qubit decoherence while in sec .
[ sec : two ] the results of ( dis)-entanglement dynamics are presented . in sec.[sec : summary ] , we summarize and discuss implication of our results .
we consider a system - bath model which is described by the total hamiltonian @xmath1 , where @xmath2 is the hamiltonian of a single or two qubits system , @xmath3 is the hamiltonian of a spin bath and @xmath4 represents the interaction between qubits and the bath
. we shall set @xmath5 but our method can be applied to a generic @xmath2 .
we shall assume that the spin bath is a spin chain characterized by the @xmath0 heisenberg model @xmath6 where @xmath7 .
it is known that the @xmath0 heisenberg model has a very rich structure.@xcite the system is ferromagnetic for @xmath8 , antiferromagnetic ( ising - type ) for @xmath9 , and critical ( xy - type ) for @xmath10 .
it also encompasses the @xmath11 model where @xmath12 .
the most general linear coupling between qubit @xmath13 and the bath can be expressed as @xmath14 , where @xmath15 and @xmath16 . here
@xmath17 characterizes the coupling of the spin qubit to the @xmath18th spin in the spin chain . for most situations ,
the more interesting cases are @xmath19 @xcite and hence we shall concentrate on negative @xmath20 . our numerical method ,
however , can be applied equally well to cases with positive @xmath20 . in our work
, both the ising coupling and isotropic heisenberg coupling will be considered . the ising coupling ( @xmath21 )
gives rise to a pure dephasing model while the isotropic heisenberg coupling ( @xmath22 ) induces both dephasing and energy relaxation .
the range of the coupling is crucial for characterizing the interaction of qubits to the spin bath
. for uniform coupling @xmath17 is independent of @xmath18 .
this is unrealistic but for uniform coupling the loschmidt echo and the entanglement dynamics can be calculated exactly by using jordan - wigner transformation when the spin bath is the @xmath11 model ( @xmath12).@xcite however , the more realistic coupling model is the local coupling model in which only the coupling to the closest spin is nonvanishing .
nonetheless , there is no analytic solutions known for this model .
the reduced dynamics is less studied but is more relevant to real experiments . in this case , if the spin bath is ferromagnetic , the qubit couples directly to the order parameter ( the magnetization ) ; while if the spin bath is antiferromagnetic or paramagnetic , the qubit does not couple to the order parameter .
hence the reduced dynamics exhibits completely different behavior in different phases . as
the local coupling model is more relevant to real experiments , in the following we shall concentrate on the local coupling model .
we now briefly outline the procedure to evaluate the reduced dynamics of the qubits and other derived quantities . for a given set of parameters
, we first employ static dmrg@xcite to find the ground state @xmath23 of the spin chain , where open boundary condition ( obc ) is used .
we assume that at @xmath24 the initial total state is a product state of the form : @xmath25 , where @xmath26 is some particular system state that we are interested in .
formally the evolution of the reduced density matrix can be obtained by evolving first the total state @xmath27 then tracing off the spin bath @xmath28 loschmidt echo and concurrence then can be evaluated from @xmath29 . in
general evolving such a state is a formidable task .
t - dmrg , however , provides a way to efficiently evolve such a state with high accuracy for a quasi - one dimensional system .
we note that the degrees of freedom of the qubits are kept exactly during the t - dmrg calculation by targeting an appropriate state .
the dimension of the truncated hilbert space is set to be @xmath30 . for short time decay simulation
we set @xmath31 in the trotter slicing while for entanglement dynamics we set @xmath32 to balance the trotter error and truncation error .
in this section , we present our results for the single qubit decoherence which is characterized by the loschmidt echo .
the loschmidt echo has been used extensively to quantify the single qubit decoherence , especially its connection to the quantum criticality of the spin baths .
@xcite loschmidt echo can be understood intuitively as follows : consider an initially disentangled total state @xmath33 , at some later time @xmath34 it will evolve into an entangled state @xmath35 .
loschmidt echo , defined as @xmath36 , clearly measure the coherence between @xmath37 and @xmath38 .
when @xmath39 the qubit is disentangled from the bath while when @xmath40 the qubit is totally entangled with the bath .
we start by noting that for numerical calculation on finite length , all dynamics will show quasi - periodic behavior .
the quasi - period is known as the revival time . since in our numerical calculation
, the spin bath is a chain of finite length , it is essential to identify the revival time for each length to avoid unphysical results due to revival . as a zeroth order approximation ,
the revival time is proportional to the length and inverse proportional to the maximum phase velocity of the spin chain . in figs .
[ fig : revival - i ] and [ fig : revival - h ] , we plot the loschmidt echo as a function of time for the case of ising and heisenberg coupling using various lengths and @xmath41 , from which the revival time of the echo can be easily identified .
we also plot the revival time as a function of the length .
one clearly observes the liner dependence of revival time on the length .
we find that for ising coupling the minimal value loschmidt echo reached is non - zero unless a very strong coupling strength is taken ( not shown here ) , while for heisenberg coupling , the loschmidt echo reaches zero if the length is longer than some @xmath41 dependent critical length .
( a)(b)(c ) @xmath42 as a function of time for different lengths and @xmath41 .
( d)revival time as a function of length .
the coupling is of ising type with @xmath43.,width=288 ] ( a)(b)(c ) @xmath42 as a function of time for different lengths and @xmath41 .
( d)revival time as a function of length .
the coupling is of heisenberg type with @xmath43.,width=288 ] to compute the single qubit decoherence , we couple the qubit to a single site of spin chain , which is taken to be the middle site of the chain to suppress boundary effects .
we tune the spin bath to different quantum phases by changing the parameter @xmath41 . in the ferromagnetic regime , a small uniform external field
is introduced during the static - dmrg calculation but is turned off during time evolution .
it is numerically checked that the numerical results reported below are insensitive to the magnitude of the applied external field .
similarly , when the bath is in the ising antiferromagnetic ground state of the xxz model , a small staggered external field is applied to lift the two - fold degeneracy in the ground state.@xcite when the spin is in the ising antiferromagnetic regime or the xy critical regime , we find that in short time the behavior of the loschmidt echo decay is gaussian , @xmath44 , where @xmath45 is the decay parameter.@xcite in fig .
[ fig : le - z ] we plot the decay parameter as a function of @xmath41 for the case of ising coupling . in the ferromagnetic regime ( @xmath46 ) , the qubit decay is completely suppressed ( @xmath47 ) .
this is a consequence of the ising coupling in which both @xmath48 and @xmath49 .
decay parameter as a function of @xmath41 for the case of ising coupling .
here @xmath43 , and length @xmath50.,width=288 ] @xmath42 as a function of time when the spin bath is in the ferromagnetic phase for the case of heisenberg coupling . here
length @xmath50 and the coupling strength @xmath43 ( upper ) or @xmath51 ( lower ) clear oscillatory behaviors are seen for both positive and negative @xmath20.,width=288 ] decay parameter as a function of @xmath41 for the case of heisenberg coupling , @xmath43 , and length @xmath50,width=288 ] clearly , the decay parameter is largest in the critical regime ( @xmath52 ) and it decreases gradually to zero as one moves into the antiferromagnetic regime ( @xmath9 ) . for single link scenario ,
the decay parameter is almost featureless within the critical regime .
our numerical results also show that if the qubit is coupled to multiple sites , decay parameter acquires a weak dependence on @xmath41 and the transition near @xmath53 becomes less sharp ( not plotted ) .
note that the decay parameter becomes sensitive to the magnitude of the small staggered field applied when the spin bath is close to the phase boundary @xmath54 .
this is due to the fact that for finite @xmath55 the barrier between two degenerate ground states is finite and approaches zero as @xmath41 approaches @xmath56 .
the ground state obtained by static dmrg includes a small mixture of the degenerate state , which is sensitive to the strength of the staggered field .
for larger @xmath41 , the barrier between two degenerate ground states increases as one moves deep into the antiferromagnetic regime . as a result
, the decay parameter becomes less sensitive to the strength of the stagger field for @xmath57 .
we now turn to the case of heisenberg coupling .
we first note that in the ferromagnetic regime , the qubit couples to the order parameter .
therefore , the qubit is effectively in an average magnetic field @xmath58 . as a consequence of the heisenberg coupling
, the qubit will precess about @xmath59 .
since magnons are generated at the same time when the qubit evolves , @xmath58 starts to deviate from @xmath60 and results in oscillations in the reduced dynamics .
[ fig : positive ] shows some typical oscillating behavior of @xmath42 in this scenario . clearly , the reduced dynamics is no longer gaussian .
therefore , we shall not mark the ferromagnetic regime in the followings . in fig .
[ fig : le - xyz ] we plot the decay parameter as a function of @xmath41 for the case of heisenberg coupling .
the overall behavior is very similar to the case of ising coupling except that the decay parameter depends weakly on @xmath41 in the critical regime .
this is different from the ising coupling case shown above but is similar to the multiple sites ising coupling case . for both ising and heisenberg coupling
we find a discontinuity in the behavior of @xmath42 at @xmath61 and a first derivative discontinuity at @xmath53 .
these discontinuities coincides with the phase boundary of the underlying spin chain .
different behavior at the @xmath62 can be traced back to the different nature of the ferromagnetic and antiferromagnetic transition and the close relation between the decoherence and the quantum criticality of the bath is clearly demonstrated .
in this section , we investigate the entanglement dynamics of two qubits that couple to the spin bath .
there are two central issues to be addressed .
the first issue is the possibility of entanglement creation via the common spin bath for a pair of initially disentangled qubits without direct interaction .
the second issue is the disentanglement dynamics of an initially entangled state . in particular
, we would like to address the issue if qubits influenced by spin baths also suffer from entanglement sudden death and if the entanglement sudden death depends on the quantum phase that qubits couple to.@xcite to characterize the entanglement , we shall use concurrence as the measurement of entanglement.@xcite for a given reduced density matrix @xmath63 , the concurrence is defined as @xmath64 , where @xmath65 are the square roots of the eigenvalues of the operator @xmath66 and @xmath67 is the complex conjugation of @xmath68 .
entanglement dynamics for an initially disentangled pair of qubits for the case of ising coupling here @xmath12 and @xmath50.,width=288 ] entanglement dynamics for an initially disentangled pair of qubits for the case of heisenberg coupling . here
@xmath12 and @xmath50.,width=288 ] it has been shown that entanglement can be created without direct interaction if two qubits interact with a common bosonic bath @xcite or a fermionic bath.@xcite of particular interest to us is the onset time of the entanglement , the strength of the induced entanglement , and the time scale where the induced decoherence eventually takes over .
these considerations are important in determining if such an induced entanglement is useful in real quantum computation .
the issue is also closely related to the proposals of induced interaction via a common bath , @xcite where the effect of induced decoherence from the same bath is usually neglected during the derivation . in fig .
[ fig : ent - c - i ] we plot the concurrence as a function of time using various coupling strength and inter qubit distance .
the coupling between qubits and the spin bath is of ising type , which gives rise to a pure dephasing model .
we assume that the coupling strength is the same for two qubits ( @xmath69 ) and the initial state is taken to be @xmath70 .
we shall set @xmath12 but similar results can ba obtained for @xmath71 .
before we discuss our findings in more details we would like to comment that if markovian approximation or uniform coupling assumption are taken then one can no longer discuss the inter - qubit distance dependence .
the relation between entanglement dynamics and inter - qubit distance , however , is gaining interest since people begin to explore the non - markovian effects of the bath .
@xcite we terminate the simulation at half of the revival time , where usually loschmidt echo reaches its minimum , to avoid the unphysical dynamics due to the revival .
we also numerically check the finite size effect by comparing the entanglement dynamics from different chain lengths .
we find that the results from different chain lengths agree with each other reasonably well .
the length of the chain mainly set an upper bound for the simulation time .
we find that for this configuration it is possible to create entanglement via the spin bath . in particular , for weaker coupling strength the induced entanglement rises more slowly but can reach a higher value ; while for stronger coupling the induced entanglement rises more rapidly .
the maximal concurrence reached , however , is lower .
this is because larger coupling strength also lead to stronger decoherence .
it is also evident from the figure that the entanglement creating rate decreases as the inter qubit distance increases , which is typical for this kind of induced interaction .
we find that for large enough @xmath20 and the smaller enough inter qubit distance the concurrence shows oscillatory behavior . in these cases , the coupling is strong enough to create concurrence oscillation but also weak enough to prevent the bath to totally disentangle the qubits .
the delicate interplay between induced decoherence and induced entanglement indicates that using such an induced entanglement for quantum computation is a tricky task .
one has to tune the coupling to be within the right window to balance the effect from each side . in fig .
[ fig : ent - c - h ] we plot the concurrence as a function of time for the case of heisenberg coupling , starting form the same initial condition .
qualitatively the behavior is similar to the case of ising coupling .
we find that the maximal entanglement that can be reached is smaller .
this is because for heisenberg coupling the loschmidt echo always decays to zero regardless the coupling strength while for ising coupling the minimal loschmidt echo value is a decreasing function of the coupling strength and is not zero .
we also find that the onset time is roughly proportional to the inter qubit distance .
this is expected as the excitation of spin chain , which mediate the entanglement generation , travels with finite phase velocity .
the time the excitation reaches the other qubit would be proportional to the inter qubit distance . however , the concurrence oscillation is absent , indicating that the induced interaction is weaker for heisenberg coupling .
we note that it is difficult to write down an exact form of the induced interaction unless markovian approximation is taken .
in general the induced interaction is time - dependent and is accompanied by a complicate decoherence effect .
it is , however , possible to perform quantum state tomography experimentally or numerically to extract kraus operators .
the kraus operators can then be used to design quantum operations without using directly the form of the induced interaction .
disentanglement dynamics for an initially disentangled pair of qubits when the bath is in the ferromagnetic phase .
the coupling is of heisenberg type , @xmath43 and @xmath50.,width=288 ] disentanglement dynamics for an initially disentangled pair of qubits when the bath is in the antiferromagnetic or the xy critical phase .
the coupling is of heisenberg type , @xmath43 , and @xmath50.,width=288 ] inverse finite disentanglement time as a function of @xmath41 . here
the coupling is the heisenberg type , @xmath43 , and @xmath50.,width=288 ] here we present our results for the disentanglement dynamics of an initially entangled state . to investigate the possibility of esd in spin bath we start from an initial state of the form @xmath72 , with an initial concurrence @xmath73 .
two qubits are set be 20 sites apart so that the decoherence of individual qubit are nearly independent and the coupling is of heisenberg type .
we first show a typical behavior of concurrence in the ferromagnetic regime in fig .
[ fig : positivecc ] . clearly , as for the reduced dynamics for a single qubit , the concurrence shows oscillatory behaviors .
hence qubits in the ferromagnetic regime do not suffer from esd and the envelop of the entanglement decays exponentially . in fig.[fig :
esd ] we plot the disentanglement dynamics of two states in the xy critical and antiferromagnetic regimes .
all of the computations start from the same initial concurrence with two set of coefficients , @xmath74 or @xmath75 , corresponding to two different initial states .
we find that in the critical regime @xmath76 both states suffers from esd .
furthermore , the entanglement dynamics of two states are identical in the critical regime , which is due to that the rotational symmetry is not broken in the critical regime and @xmath77 is equivalent to @xmath78 by @xmath79 symmetry along the quantization axis . in both ferromagnetic and antiferromagnetic phase , where the rotational symmetry is broken ,
we find that the entanglement dynamics for these two states starts to deviate from each other . in most of the antiferromagnetic regime both states do not suffer from esd .
when @xmath41 is close to the phase boundary , however , both of two states that correspond to @xmath74 and @xmath75 suffer from esd and have slightly different disentanglement time . in fig .
[ fig : esd - phase ] we plot the inverse of disentanglement time , which is defined as the time when concurrence becomes zero , as a function of @xmath41 .
starting from @xmath61 , it shows monotonic decrease . across the phase boundary , @xmath53 ,
the inverse of disentanglement time develops a small bump and persists into the antiferromagentic phase and finally decreases to zero around @xmath80 .
the existence of finite region with finite disentanglement time in the antiferromagnetic regime is due to that when @xmath41 approaches @xmath56 , the barrier between two degenerate ground states approaches zero . for finite @xmath55 and finite value of @xmath81 , the ground state obtained by static dmrg
includes a small mixture of the degenerate state so that it resembles the xy critical state and results in finite disentanglement time . in the thermodynamic limit ( @xmath82 ) , the region with finite disentanglement time in the antiferromagnetic regime shrinks down , resulting discontinuity at the phase boundary @xmath53 .
the overall behavior found in the above is different from those reported in the early work@xcite in which two states investigated in the above posses different entanglement dynamics and only one of them suffer from esd .
the difference is due to that the model adopted in ref .
@xcite includes the effect of spontaneous decay , breaking the symmetry between @xmath83 and @xmath84 ; while for spin bath in xy critical regime such symmetry breaking is absent .
it is important to note that there exits a subregime , roughly around @xmath85 , in which entanglement shows revival after some dark period . note that the entanglement revival after some dark period was also reported in ref .
@xcite where a photonic multimode vacuum bath is assumed and the revival is attributed to the two photon decay . in our work ,
the origin of the revival is less clear .
we believe that the existence of such sub - regime is due to the competition between entanglement decay and loschmidt echo decay . within this subregime the loschmidt echo decay more slowly , giving the system a chance to revive after the first esd .
in summary , the decoherence and ( dis)-entanglement dynamics induced by spin baths are investigated non - perturbatively by using t - dmrg . for both pure dephasing model ( ising coupling ) and general decoherence model ( heisenberg coupling )
we calculate the short time decay parameter of the loschmidt echo .
we find that in both cases the decay parameter is closely related to the phase of the underlying spin chain . in the ferromagnetic regime ,
the reduced dynamics shows oscillatory behavior while in the xy critical and antiferromagnetic regimes , the decay parameter shows a first derivative discontinuity at @xmath86 .
we evaluate the entanglement dynamics of a pair of initially disentangled qubits which are close to each other .
we demonstrate that it is possible to induced entanglement via their common interaction with the spin bath .
the competition between induced decoherence and entanglement can be easily seen in the coupling strength dependent behavior of the entanglement onset time , growth rate , and the maximal entanglement reached .
finally we investigate the disentanglement dynamics of a pair initially entangled qubits which are far from each other .
for the two initial states we studied , we find that their disentanglement dynamics are identical and suffer from esd in the critical regime .
their disentanglement dynamics begin to deviate from each other in both the ferromagnetic and the antiferromagnetic regime .
they no longer suffer from esd in the ferromagnetic regime but still suffer from esd if the chain is near the antiferromagnetic transition .
it is shown that the inverse of finite disentanglement time has a close relation to the phase of the spin bath and shows monotonic decrease behavior as one moves into the antiferromagnetic regime .
|
the reduced dynamics of a single or two qubits coupled to an interacting quantum spin bath modeled by a xxz spin chain is investigated . by using the method of time - dependent density matrix renormalization group ( t - dmrg ) , we go beyond the uniform coupling central spin model and evaluate nonperturbatively the induced decoherence and entanglement .
it is shown that both decoherence and entanglement strongly depend on the phase of the underlying spin bath .
we show that in general , spin baths can induce entanglement for an initially disentangled pair of qubits .
furthermore , when the spin bath is in the ferromagnetic phase , because qubits directly couple to the order parameter , the reduced dynamics shows oscillatory type behavior . on the other hand , only for paramagnetic and antiferromagnetic phases , initially entangled states suffer from the entanglement sudden death . by calculating concurrence , the finite disentanglement time is mapped out for all phases in the phase diagram of the spin bath .
|
when the temperature @xmath5 is high compared with the typical momentum scale but well below the planck scale , all the n - graviton thermal green functions can be computed in the one - loop approximation using the hard thermal loop expansion .
there have been many investigations where this approach has been employed @xcite . an important property which is now well established is the gauge invariance of the leading high temperature contributions of all n - graviton thermal green functions .
the explicit from of these contributions can be obtained using the equivalence which exists between the formalism of boltzmann transport equation and the high temperature limit of the thermal green functions in field theory@xcite . using this approach ( which is explicitly gauge invariant )
one can easily show that the leading part of all n - point one - loop thermal green functions is proportional to @xmath0 @xcite .
these results have also been obtained by standard feynman diagrammatic calculation in the feynman - dedonder gauge for the one- and two - graviton functions @xcite as well as for the three - graviton function @xcite .
one of the interesting physical applications of the one- and two - graviton functions is the study of the dispersion relations@xcite which follow from linear response theory@xcite . since the relevant physical quantities are obtained from the _ poles _ of the propagator , it is important to verify the gauge independence of this procedure .
while this is automatically fulfilled by the leading high temperature contributions , the inherently gauge dependent _ sub - leading contributions _ to the thermal green functions require a more detailed investigation .
a similar situation occurs in the case the yang - mills theory where it is known that a gauge independent set of dispersion relations can be obtained from the gauge dependent thermal two - gluon function @xcite .
as far as we know , in contrast with the case of the yang - mills theory , a gauge independence proof of the dispersion relations in quantum gravity beyond the leading order is still missing .
the purpose of the present paper is to investigate this problem in the case of gravity using the standard feynman diagrammatic approach .
we will compute the 1- and 2-graviton functions up to sub - leading contributions , in a class of general gauges .
( some intrinsically gauge independent sub - leading contributions have been considered before using the theory of a scalar field in interaction with gravity@xcite . )
we employ the _ imaginary time formalism_@xcite and express the one - loop thermal green functions in terms of on - shell forward scattering amplitudes ( the `` barton amplitude'')@xcite , properly generalized in order to account for the quadratic denominators which arises in the free graviton propagator when a general gauge fixing term is employed @xcite ( see also appendix [ appa ] ) .
this approach enable us to explore some of the general properties of the _ exact _ graviton self - energy without having to carry out explicitly the non - trivial spatial momentum integrations .
it is also much more straightforward to perform the hard thermal loop expansion when we start from the forward scattering amplitudes . using this approach
we were able to obtain explicit results for the @xmath1 and logarithmic contributions to the graviton self - energy . unlike the leading high temperature terms , for which the gauge independence is confirmed by our calculation , the sub - leading contributions are gauge dependent . these contributions will be employed in the study of the dispersion relations for the transverse and traceless gravitational modes@xcite . this paper is organized as follows : in sect .
[ sect2 ] we present the lagrangian and the basic definition of the graviton field from which the feynman rules are derived .
we also discuss the identities which follows from the gauge invariance of the theory . in sect .
[ sect3 ] the main results of the calculation of the one- and two - graviton functions up to the logarithmic contributions is presented .
a very compact expression for the @xmath1 contribution is obtained and we verify that the logarithmic contribution is proportional to the ultraviolet pole part of the @xmath3 two - graviton function .
we also derive the general transformation of the two - graviton function under a change of graviton representation . in sect .
[ sect4 ] we verify explicitly that the gauge dependence of the sub - leading @xmath1 contributions to the two - graviton function is effectively canceled in the dispersion relations .
we then present the solutions of these equations , which include corrections of order @xmath1 to the leading @xmath0 contributions , describing the physical modes for the propagation of waves in a graviton plasma .
the feynman rules for the graviton propagator and self - interactions vertices are obtained from the following underlying lagrangian @xmath6 where @xmath7 is the ricci scalar , g is the newton constant and the parameter @xmath8 defines a family of gauges .
( @xmath9 is the feynman gauge and @xmath10 is the landau gauge ) .
the quantities @xmath11 and @xmath12 are the _ ghost fields _ and the function @xmath13 is the infinitesimal generator of coordinate ( gauge ) transformations @xmath14 the calculations in quantum gravity are conveniently performed using the _ graviton field _ @xmath15 defined in terms of the tensor @xmath16 as @xmath17 where @xmath18 is the minkowski metric . the feynman rules can be obtained in a straightforward way substituting ( [ field ] ) into ( [ lagrangian ] ) and performing a perturbative expansion in @xmath19 .
the 0th order terms are quadratic in the graviton field and yield the following expression for the graviton propagator @xmath20 in the appendix [ appb ] we give all the other relevant feynman rules employed in this work .
the choice of the graviton field parametrization given by ( [ field ] ) restricts the gauge parameter dependence only to the propagator ( [ propag ] ) , since in this case the second term in ( [ lagrangian ] ) is exactly quadratic in the graviton field @xmath21 .
this is similar to the general _ linear gauges _ in yang - mills theories .
therefore , the gauge dependence of the green functions computed from these feynman rules can be traced back to eq .
( [ propag ] ) .
the leading high temperature contributions of all one - particle irreducible thermal green functions are related to each other through tree - like ward identities in the same way as the basic tree vertices @xcite .
these hard thermal loop identities have been verified for both yang - mills theories and gravity and generalized to any gauge theory whose generators form a closed algebra@xcite . for our present purposes
it will be sufficient to consider the identity involving the two - point function .
a simple example is provided by the following tree ward identity , arising from the invariance of the pure einstein action @xmath22 under the transformation given by equation ( [ transf ] ) , @xmath23 where @xmath24 is the tensor generated from the transformation of the graviton field under ( [ transf ] ) and @xmath25 comes from the quadratic term in the action without the gauge fixing term ( it is the inverse of the propagator in the limit @xmath26 ) . the tree - like identity which holds for
the high temperature limit of the two - graviton function would be identical to eq .
( [ x0s0 ] ) if the one - graviton function ( the tadpole ) , shown in the diagrams in fig .
1 , were zero .
the modification introduced by the tadpole changes the right hand side of eq .
( [ x0s0 ] ) to a non zero quantity when @xmath27 is replaced by the leading high temperature contribution of @xmath28 , given by the diagrams in ( @xmath29 is a time - like normalized four - vector representing the local rest frame of the plasma ) .
this contrasts with the analogous situation in the case of yang - mills theories where the anti - symmetry of the group structure constants trivially makes the tadpole to vanish . as a consequence of the non - vanishing tadpole
, the general brst identities will not hold for the _ exact _ finite temperature graviton self - energy .
however , as we will see in the next section , the tadpole diagrams can be computed _ exactly _ , yielding a result proportional to @xmath0 .
therefore , if we split @xmath28 as @xmath30 the brst identities derived in the appendix [ appc ] will hold for the sub - leading contributions @xmath31 , so that the following identity is satisfied @xmath32 this identity is analogous to @xmath33 , where @xmath34 is the exact gluon self - energy@xcite .
since all the gauge parameter dependence is restricted to the sub - leading contributions , these identities have an important rle in the cancellation of the gauge dependence in the dispersion relations .
in this section we will present the details of the calculation of the one- and two - graviton functions .
let us first consider the contributions from the two tadpole diagrams in figure 1 .
the most involved diagram is the one shown in figure 1a , since both the 3-graviton vertex and the general gauge propagator are involved . using eq .
( [ 3grav ] ) and the propagator ( [ propag ] ) , the straightforward contraction of indices yields a result which is independent of the parameter @xmath35 .
therefore , the resulting expression is identical to what is obtained in the feynman - dedonder gauge involving only the usual _
quadratic denominators_. using the eq .
( [ genint ] ) in the simple case when @xmath36 and @xmath37 the following result for the one - graviton function is readily obtained @xmath38 where @xmath39 denotes the angular integral and the four vector @xmath40 is on - shell with components given by @xmath41 .
the diagrams contributing to @xmath28 are shown in figure 2 .
the contributions associated with each of these diagrams will involve integrals like the one shown in eq .
( [ genint6 ] ) . from the structure of the graviton propagator given by eq .
( [ propag ] ) we can see that the diagram in figure 2a is such that each of its terms will involve integrals with @xmath42 , while in the case of the diagram shown in figure 2b , all the corresponding integrals have the form of the first term of eq .
( [ genint6 ] ) with @xmath37 and @xmath43 . in the case of the ghost loop diagram shown in figure 2c
, all the terms will involve integrals with @xmath44 .
let us first consider the leading high temperature behavior of these integrals . in this limit
, we can perform a hard thermal loop expansion of the integrand such that the terms with @xmath45 will all be sub - leading . for the terms with @xmath46 we use expansions like @xmath47 the case @xmath48 ( from the diagram in figure 2b ) is similar to the tadpole diagram giving an exact @xmath0 contribution . in this way , we obtain the following result for the leading behavior of the graviton self - energy @xmath49 it is worth mentioning that though a nave power counting would allow for a gauge parameter dependence from the third term in the second line of eq .
( [ propag ] ) , the final result ( [ pit4 ] ) is gauge independent as one would expect on more general grounds @xcite .
combining the eqs .
( [ break ] ) , ( [ thooft ] ) and ( [ pit4 ] ) we obtain the following identity for the exact self - energy @xmath50 where in the last term we have used ( [ tadpole ] ) . since the integrand in the right hand side of the above expression is an elementary expression without denominators , the same should be true for its left hand side , up to terms which would vanish after integration .
our calculation shows that , in fact , the expression obtained from the diagrams in figure 2 is such that the exact integrand of @xmath51 does not involve any denominators , being identical to the integrand in the right hand side of ( [ thooft1 ] ) .
we have extended the hard thermal loop expansion in order to obtain the @xmath1 and the _ logarithmic contributions _ , which are yielded respectively by the terms of degree @xmath52 and @xmath53 in @xmath54 ( terms of degree @xmath55 in @xmath54 are absent due to the symmetry @xmath56 ) from the expansion of the integrand in expressions like ( [ genint6 ] ) in the region of large values of @xmath54 .
after a long computation we have been able to find the following compact expression for the @xmath1 contribution @xmath57\right\ } , \end{array}\ ] ] where @xmath58\ ] ] and @xmath59.\ ] ] using eq .
( [ x0s0 ] ) and the structure of ( [ pit2 ] ) we immediately conclude that @xmath60 this result can be understood in the context of the brst identities , using the results of appendix [ appc ] .
it is remarkable that though this @xmath1 contribution is gauge dependent , it is transversal to @xmath61 .
it is straightforward to obtain the explicit results for the angular integrals in eqs .
( [ pit4 ] ) , ( [ if ] ) and ( [ ig ] ) in terms of a tensor basis such as the one shown in the table [ tab1 ] . using the following decomposition @xmath62 where @xmath63 is a function of degree 2 or 0 in @xmath64 respectively for the leading @xmath0 or the @xmath1 contributions ,
the coefficients @xmath65 are obtained contracting both sides of eq .
( [ decomposition ] ) with each of the 14 tensors of table [ tab1 ] .
the solution of the resulting linear system of 14 equations is given in terms of integrals like @xmath66 , which can be easily evaluated .
in the case of the logarithmic contributions the resulting angular integrals @xmath39 can all be parametrized in a _
lorentz covariant _ way in terms of the 5 tensors @xmath67 , @xmath68 , @xmath69 , @xmath70 , @xmath71 .
the result can be expressed in terms of the @xmath3 graviton self - energy @xcite in the following way @xmath72 where @xmath73 is the residue of the ultraviolet divergent @xmath3 contribution which is obtained from the calculation in @xmath74 dimensions .
the fact that both the @xmath2 and the ultraviolet divergent contributions have the same structure has been also verified for the two and four - point functions in qed @xcite and for the two- and three - point functions in yang - mills theories @xcite .
these results are special examples of the rather general arguments presented in @xcite . from the results for the thermal one- and two graviton functions we can write the following expression for the _ thermal effective action _ @xmath75=\gamma_{\mu\nu}\,h^{\mu\nu}(0)+ \int { \rm d}^4 k\ , h^{\mu_1\nu_1}(k ) \pi_{\mu_1\nu_1\,\mu_2\nu_2}(k ) h^{\mu_2\nu_2}(-k ) + \cdots.\ ] ] here we will use eq . ( [ stherm ] ) in order to derive the expressions for the new one- and two - graviton functions which arise when one uses the graviton representation @xmath76 the corresponding expressions will be employed in the analysis performed in the next section .
using eqs .
( [ field ] ) and ( [ field1 ] ) one obtains the following relation for the graviton fields in the two representations @xmath77 inserting eq .
( [ representation ] ) into eq .
( [ stherm ] ) and using the traceless property of @xmath78 [ cf .
( [ tadpole ] ) ] , we obtain @xmath79=\check\gamma_{\mu\nu}\,\delta g^{\mu\nu}(0)+ \int { \rm d}^4 k\ , \delta g^{\mu_1\nu_1}(k ) \,\check \pi_{\mu_1\nu_1\,\mu_2\nu_2}(k)\ , \delta g^{\mu_2\nu_2}(-k ) + \cdots , \ ] ] where @xmath80 and @xmath81 we remark that while the derivation of eq .
( [ pipicheck ] ) is rather simple and general , a direct calculation of @xmath82 , on the other hand , would involve the manipulation of more complicated feynman rules where the gauge fixing term from eq .
( [ lagrangian ] ) would contribute to all the n - graviton vertices .
the thermal graviton self - energy has been employed in order to investigate the propagation of gravitational waves in a plasma @xcite .
this can be done studying the poles of the full graviton propagator ( dispersion relations ) which is obtained from the effective action @xmath83=s_g[g ] + s_{fix}[g ] + s_{term}[g]\ ] ] where @xmath84 $ ] is the einstein action , @xmath85 $ ] is the gauge fixing term and @xmath86 $ ] is given by eq .
( [ stherm ] ) . in this section
we shall apply the results for the graviton self - energy up to the sub - leading @xmath1 contributions in order to investigate the gauge dependence of the dispersion relations .
since the tadpole contribution to @xmath86 $ ] yield a non - zero energy - momentum tensor in the einstein equation @xmath87}{\delta g_{\mu\nu}}=0,\ ] ] a self - consistent calculation of the full graviton propagator has to take into account a _ curved background _ so that @xmath88 where @xmath89 is the solution of the einstein equation ( [ einstein ] ) and @xmath90 is the _
metric fluctuation_. from the corresponding second order variation of the effective action @xmath91 = -\frac 1 2 \int { \rm d}^4 x \sqrt{-g^{(0)}}\ , \delta g_{\mu_1\nu_1}p^{\mu_1\nu_1\,\mu_2\nu_2 } \delta g_{\mu_2\nu_2},\ ] ] the graviton propagator can be obtained taking the inverse of @xmath92 .
the contributions to @xmath92 from the first two terms in ( [ effective ] ) are well known @xcite .
they involve components of the riemann and ricci tensors and the scalar curvature . restricting the analysis to a metric background which is _ conformally flat _ , the components of riemann tensor can be expressed only in terms of the ricci tensor and the scalar curvature . since eq .
( [ tadpole ] ) yields a traceless energy - momentum tensor , the einstein equation ( [ einstein ] ) ( with vanishing cosmological constant ) implies that the scalar curvature is zero and that the ricci tensor is proportional to eq .
( [ tadpole ] ) . using _
geodesic normal coordinates _ the thermal contributions to @xmath92 can be obtained from eq .
( [ stherm1 ] ) after a straightforward tensor algebra one obtains the following expression @xmath93 \\ { } & { } & \biggl .
+ \left({\rm symmetrization\;\;under\;\ ; \mu_1 \leftrightarrow \nu_1}\right)\biggr\ } - 16\pi g\check\pi ^{\mu_1 \nu_1 \,\mu_2 \nu_2 } \left(k , u\right ) \end{array},\ ] ] where @xmath94 is given by eq .
( [ pipicheck ] ) with the leading and the sub - leading high - temperature contributions from @xmath95 given respectively by eqs .
( [ pit4 ] ) and ( [ pit2 ] ) .
because of the coordinate invariance of the problem we have to impose physical constraints on the metric fluctuations .
the imposition that the spin one and spin zero degrees of freedom do not propagate constraints the metric perturbations @xmath90 to be _ transverse _ and _ traceless _
, respectively@xcite .
these conditions imply that we only have to consider the transverse and traceless components of @xmath96 in the linear response equation @xmath97 an explicit basis of tt - tensors can be found imposing the tt conditions on a general linear combination such as the one on the right hand side of eq .
( [ decomposition ] ) .
it is also convenient to choose these tensors as being idempotent and orthogonal to each other .
this leads to the following set of tt - tensors @xmath98 where the coefficients @xmath99 are given in the table [ tab2 ] .
this result is in agreement with the one obtained by rebhan in reference @xcite ( except for a small misprint in the 9th line of the 2nd row in table 2 ) . as a simple checkup of this result we note
that at zero temperature there is only one tt - tensor given by @xmath100 so that in the table [ tab2 ] the lines @xmath101 , @xmath102 , @xmath103 , @xmath104 and @xmath105 ( which gives the coefficients of the lorentz covariant tensors in the table [ tab1 ] ) must add to a lorentz scalar ( or a pure number ) and all the other lines must add to zero .
once we know a certain set of coefficients @xmath106 in the basis given by table [ tab1 ] and the explicit form of the tt - tensors given by eq .
( [ tttensors ] ) a straightforward calculation gives the following result for the coefficients in the basis of tt - tensors @xmath107 it is interesting to note that the sub - leading contributions to the graviton self - energy are independent of the graviton representation .
this property is satisfied because these contributions to @xmath94 and @xmath95 have the _ same _ tt components .
indeed , we see from eq .
( [ pipicheck ] ) that apart from the tadpole contributions which have an exact @xmath0 behavior .
the terms involving traces of @xmath95 are either proportional to @xmath108 or @xmath109 or both .
such terms have no components along any of the first 3 tensors of table [ tab1 ] , so that they give no contribution to any of the coefficients @xmath106 , @xmath110 which appear in eqs .
( [ ttcoeffs ] ) .
we have now all the basic quantities which are needed in order to express @xmath92 in the basis of tt - tensors as follows @xmath111 the inverse @xmath112 can be determined from the relation @xmath113 using the transversality and idempotence of @xmath114 as well as the identities @xmath115 we obtain the following result @xmath116 we can now investigate the poles of the tt components of the propagator from the solution of the equations @xmath117 . using the eqs .
( [ ttcoeffs ] ) with @xmath118 , @xmath119 and @xmath120 determined from the decomposition of ( [ pp ] ) in the basis of table [ tab1 ] , the equations associated with the modes @xmath121 , @xmath122 and @xmath123 can be written respectively in the form @xmath124{lcl } { k^2 } & = & { \displaystyle{\frac{16\pi\ , g\rho } { 1-\displaystyle{{{\displaystyle{\frac{64\pi}{15 } } } \ , { g\ , t^2 } } } \left({1- \xi}\right ) } } } \left [ \left ( \displaystyle{{\displaystyle{\frac{5}{9 } } } } + \displaystyle{{\displaystyle{\frac{1}{2}}}}\,{r}^{4 } l-\displaystyle{{\displaystyle{\frac{1}{6}}}}\,{r}^{2 } \right)+ { \displaystyle{\frac{8\ , k^2}{\pi^2 t^2}}}\left(8\,{r}^{2}\,l + { \displaystyle{\frac{1}{2}}}{r}^4 l - { \displaystyle{\frac{1}{6}}}\,{r}^{2}-4\right ) \right ] \\ { } & { } & \\
{ k^2 } & = & { \displaystyle{\frac{16\pi\ , g\rho } { 1+\displaystyle{{{\displaystyle{\frac{64\pi}{15 } } } \ , { g\ , t^2 } } } \left({1- \xi}\right ) } } } \left [ \left(\displaystyle{{\displaystyle{\frac{2}{9 } } } } - 2\,{r}^{4 } l + \displaystyle{{\displaystyle{\frac{2}{3}}}}\,{r}^{2 } + { { \displaystyle{\frac{10}{9}}}}\ , { \displaystyle{\frac{1}{{r}^{2}}}}\right)+{\displaystyle{\frac{8\ , k^2}{\pi^2 t^2}}}\left ( -5\,{r}^{2}l\,-2\,l\,{r}^{4}+{\displaystyle{\frac{2}{3}}}\,{r}^{2}\right ) \right ] \\ { } & { } & \\ { k^2 } & = & { \displaystyle{\frac{16\pi\ , g\rho } { 1-\displaystyle{{{\displaystyle{\frac{64\pi}{15 } } } \ , { g\ , t^2 } } } \left({1- \xi}\right ) } } } \left[\left ( { \displaystyle{{\displaystyle{\frac{8}{9 } } } } } + 3\,{r}^{4 } l -{r}^{2 } + \displaystyle{{{\displaystyle{\frac{28}{27}}}}\,{\displaystyle{\frac{1}{{r}^{2}}}}}\right ) + { \displaystyle{\frac{8\ , k^2}{\pi^2 t^2}}}\left(1 - 6\,{r}^{2}l\ , + 3\,l\,{r}^{4 } - { r}^{2}\right)\right ] \end{array } \right . , \ ] ] where @xmath125 the eqs .
( [ modes ] ) reduces to the eqs .
( 6.2 ) of reference @xcite in the special case when the sub - leading terms proportional to @xmath126 or @xmath127 are neglected . in view of the constraints imposed by the important condition ( [ x0pit2 ] ) , the gauge dependent denominators in eqs .
( [ modes ] ) have a very simple structure . since we assume that @xmath128 , the momentum - independent denominators can be expanded perturbatively .
we thus see that all the gauge dependent @xmath1 sub - leading contributions to the dispersion relations give effectively corrections of order @xmath129 , which are of the same order as the two - loop contributions which we have disregarded .
hence , we conclude that to one - loop order , the dispersion relations are effectively gauge independent
. the solution of the one - loop dispersion relations can therefore be obtained from eqs .
( [ modes ] ) , by setting the denominators equal to one .
these solutions have been obtained in the reference @xcite in the leading high temperature approximation . in order to illustrate the magnitude of the @xmath1 sub - leading contributions
let us consider the solution of eqs .
( [ modes ] ) for real values of @xmath130 and @xmath131 .
this corresponds to the propagation of waves supported by the graviton plasma . in the figure 3 , where @xmath132 and @xmath133 , we show the numerical solutions corresponding to the modes @xmath121 , @xmath122 and @xmath123 and compare the leading results with the contributions which include the sub - leading @xmath1 corrections .
we can see from these diagrams that for all tt - modes , the dispersion curves begin at a common plasma frequency @xmath134 and become asymptotically parallel to the light cone .
the behavior of the dispersion relation can be determined analytically in the limiting cases of very small and very large momenta .
when @xmath135 , the common form of the dispersion relations is given by @xmath136 for large momenta such that @xmath137 , the asymptotic forms of the dispersion relations become respectively @xmath138 the small wavelength limit given by the above relations can be understood by noticing that in this case one probes the plasma at small distances , where the medium effects on the free dispersion relations are relatively unimportant . on the other hand , for long wavelengths eq .
( [ plasmaf ] ) gives a substantial modification of the free dispersion relation due to the collective phenomena in the plasma .
however , this modification is sensitive to the curvature of space which we have neglected in our analysis .
( such an effect yields corrections of magnitude @xmath139 , which are formally of the same order as the two - loop contributions ) .
this is an interesting issue which deserves further study . .
[ [ appa ] ] here we show explicitly how to extend the method of barton amplitudes in order to account for the contributions which arises from the quadratic denominators in the general gauge free propagator .
we illustrate this technique by considering the following integral @xmath140,\ ] ] where @xmath141 , @xmath142 and .
this is the most general kind of integral which contributes to the two - point function when the imaginary time formalism is employed .
the generalization to higher point functions is straightforward .
the numerator @xmath143 comes from the graviton vertices and from the numerator of the free propagator . since the integration in ( [ genint ] ) is over all values of @xmath144 , it is more convenient to make the change of variables @xmath145 in all the terms@xmath146 so that @xmath147.\ ] ] factorizing the denominators in ( [ genint2 ] ) we can write @xmath148 \end{array}.\ ] ] the @xmath149 integration is now readily performed using the cauchy theorem and closing the contour in the right hand side plane where the only poles are located at @xmath150 and @xmath151 ( @xmath130 is a pure imaginary quantity at this stage of the calculation ) . in this way we obtain
@xmath152 performing the change of variables @xmath153 in the second term of ( [ genint4 ] ) we can write @xmath154 finally , using the property @xmath155 and the symmetry @xmath56 we obtain @xmath156 the special case when @xmath44 gives the known result @xmath157_{q\cdot q = 0},\ ] ] where the expression inside the bracket is a typical contribution to the on - shell forward scattering amplitude . although the derivatives in the general expression ( [ genint6 ] ) makes it much more difficult to be handled , it is straightforward to deal with such kind of expressions using a _
computer algebra program_. [ [ appb ] ] the feynman rules are obtained inserting eq .
( [ field ] ) and the corresponding perturbative expansion of the inverse @xmath158 into eq .
( [ lagrangian ] ) .
the contributions of order 0 in @xmath19 yields the graviton propagator given by eq .
( [ propag ] ) . the third term in eq .
( [ lagrangian ] ) yields the following expressions for the ghost propagator and the graviton - ghost - ghost vertex @xmath159 @xmath160.\ ] ] all the graviton self - couplings are generated only from the first term in eq .
( [ lagrangian ] ) .
the corresponding feynman rules for the three- and four - graviton couplings are given respectively by the following expressions @xmath161\\ { } & { } & + { \rm permutations}\;\;{\rm of}\;\ ; ( k_1,{\mu_1},{\nu_1}),\;\;(k_2,{\mu_2},{\nu_2}),\;\;(k_3,{\mu_3},{\nu_3})\;\ ; , \end{array}\ ] ] @xmath162 \\ { } & { } & + { \rm permutations}\;\ ; { \rm of}\;\;(k_1,{\mu_1},{\nu_1}),\;\;(k_2,{\mu_2},{\nu_2}),\;\;(k_3,{\mu_3 } , { \nu_3}),\;\ ; ( k_4,{\mu_4},{\mu_4})\;\;. \end{array}\ ] ] as usual , we have energy - momentum conservation at the vertices , where all momenta are defined to be inwards .
the imaginary time formalism at finite temperature follows closely the corresponding formalism at @xmath3 .
consequently , the ' t hooft identities at finite @xmath5 would be similar to the ones at @xmath3 , were it not for the presence of 1-particle tadpole contributions ( such terms vanish at @xmath3 in the dimensional regularization scheme ) .
however , since the tadpole terms are proportional to @xmath0 , they do not affect the identities involving the sub - leading contributions . to derive these , we start from the action @xmath163 here @xmath164 denotes the sub - leading contributions to the graviton 2-point function and @xmath165 represents the tensor generated by a gauge transformation of the graviton field which is given to lowest order by eq .
( [ x0 ] ) in the momentum space .
@xmath166 is an external source , @xmath167 represents the ghost field and @xmath168 stand for terms which are not relevant for our purpose .
the ' t hooft identity involving the graviton self - energy function is a consequence of the brst invariance of the action @xmath169 : @xmath170 to lowest order , eq .
( [ brst ] ) is equivalent to the relation eq .
( [ x0s0 ] ) . in general , eq .
( [ brst ] ) implies the generalized ' t hooft identity @xmath171 which can be written to second order as @xmath172 using eq .
( [ x0s0 ] ) , we see that ( [ genthooft2 ] ) leads immediately to the ' t hooft identity ( [ thooft ] ) . in order to derive eq .
( [ x0pit2 ] ) , we shall need to evaluate the tensor @xmath173 which appears in eq .
( [ genthooft2 ] ) .
this tensor may be represented by the diagram shown in fig .
4 , where the ghost - graviton - source vertex is given in the appendix a of ref . @xcite . using the forward scattering amplitude method , we obtain the following structure for the @xmath1 contributions to @xmath173 @xmath174 using this form and eq .
( [ x0s0 ] ) , it is clear that @xmath175 is orthogonal to @xmath176 , so that the relation ( [ x0pit2 ] ) follows at once from the identity ( [ genthooft2 ] ) .
u. kraemmer and a. rebhan , phys .
lett 67 , 793 ( 1991 ) ; h. nachbagauer , a. k. rebhan and d. j. schwarz , phys . rev .
d * 53 * , 882 ( 1996 ) ; h. nachbagauer , a. k. rebhan , d. j. schwarz , phys . rev .
d * 53 * 5468 ( 1996 ) .
|
we study the graviton self - energy function in a general gauge , using a hard thermal loop expansion which includes terms proportional to @xmath0 , @xmath1 and @xmath2 .
we verify explicitly the gauge independence of the leading @xmath0 term and obtain a compact expression for the sub - leading @xmath1 contribution .
it is shown that the logarithmic term has the same structure as the ultraviolet pole part of the @xmath3 self - energy function .
we argue that the gauge - dependent part of the @xmath1 contribution is effectively canceled in the dispersion relations of the graviton plasma , and present the solutions of these equations . #
|
topological superconductors are analogous to topological insulators:@xcite both combine an excitation gap in the bulk with gapless states at the surface , without localization by disorder as long as time - reversal symmetry is preserved . however , the nature of the surface excitations is entirely different : in a topological insulator these are dirac fermions , relativistic electrons or holes of charge @xmath8 , while a topological superconductor has _
charge - neutral _ majorana fermions on its surface .
a transport experiment that aims to detect the majorana surface states can not be as routine as electrical conduction the direct analogue for majorana fermions of the _ electrical _ conductance of dirac fermions is the _ thermal _ conductance @xmath0 .
the challenge of low - temperature thermal measurements is one reason why majorana surface states have not yet been detected in a transport experiment on candidate materials for topological superconductivity.@xcite there exists a purely electrical alternative to thermal detection of majorana fermions.@xcite particle - hole symmetry enforces that a majorana fermion at the fermi level is an equal - weight electron - hole superposition , so while the average charge is zero , the charge fluctuations have a quantized variance of @xmath9 quantum fluctuations of the charge can be detected electrically in a shot noise measurement , and for a single fully transmitted majorana mode these produce a quantized shot noise power @xmath2 of @xmath10 per @xmath11 of applied bias.@xcite ( the factor @xmath12 reminds us that a majorana fermion is `` half a dirac fermion '' . ) is obtained by applying a temperature difference @xmath1 , @xmath13 between the contacts and measuring the resulting heat current @xmath14 . for the shot noise measurement one would bias contact @xmath15 at voltage @xmath16 , while keeping the superconductor and @xmath17 grounded .
the electrical current @xmath18 into contact @xmath17 fluctuates with noise power @xmath2 . both @xmath19 and @xmath20
are governed by the majorana surface states of the topological superconductor . ] in a two - dimensional ( 2d ) topological superconductor , studied in refs . , there is only a single majorana edge mode , but a three - dimensional ( 3d ) topological superconductor has a large number @xmath21 of majorana surface modes connecting a pair of metal contacts ( see fig .
[ fig_layout ] ) . in the absence of inter - mode coupling
this would give a shot noise power of @xmath22 for a mode - averaged transmission probability @xmath23 . because the thermal conductance equals@xcite @xmath24 with @xmath25 the lorenz number , uncoupled majorana modes have a one - to - one relationship between shot noise and thermal conduction .
we would expect this relationship to break down as a result of inter - mode scattering : a pair of coupled majorana modes is equivalent to a single dirac fermion mode , which can be in an eigenstate of charge at @xmath26 .
the thermal conductance would not be affected , as long as @xmath23 remains the same , but @xmath2 would be reduced .
much to our surprise , we discovered in numerical simulations of a 3d topological superconductor that @xmath27 with high accuracy .
this is remarkable even in the absence of any disorder , since the modes at top and bottom surfaces are coupled when they reach the metal contact .
we have found that it is the combination of electron - hole and time - reversal symmetry that preserves the relationship between electrical and thermal conduction in a 3d topological superconductor , provided the conversion from majorana to dirac fermions at the metal electrode is local in space .
the general argument is presented in secs .
[ conductionformulas ] and [ pehtsym ] .
the implication is that the shot noise power has a universal limit @xmath28 for a surface of aspect ratio @xmath6 , with corrections from poor coupling to the metal contacts that we calculate in secs .
[ from3dto2d ] and [ sec_gthermal ] . a numerical test of our analytical predictions for a model hamiltonian of a 3d topological superconductor
is given in sec .
[ sec_3d ] .
we conclude in sec .
[ conclude ] with a discussion in the context of the wiedemann - franz relation between electrical and thermal conduction.@xcite
we consider the geometry of fig .
[ fig_layout ] , a superconductor @xmath29 connecting two normal - metal contacts @xmath15 and @xmath17 .
the superconductor is topologically nontrivial , with a gapped bulk and a gapless surface .
we compare two transport properties , one thermal and one electrical , both sensitive to the surface states . for thermal transport
we take the two - terminal thermal conductance @xmath30 giving the heat current @xmath14 flowing from contact @xmath15 at temperature @xmath13 to contact @xmath17 at temperature @xmath1 , in linear response for @xmath31 .
the superconductor is a thermal insulator in the bulk , because of the excitation gap , but a thermal conductor on the surface , so @xmath0 measures heat conduction along the surface . for electrical transport both contacts
are kept at the same temperature @xmath1 .
contact @xmath15 is biased at voltage @xmath16 relative to ground , while contact @xmath17 as well as the superconductor are grounded .
most of the charge current @xmath32 injected into the superconductor at @xmath15 is short - circuited to ground via the bulk , which is an ideal electrical conductor . at the remote contact @xmath17
a fluctuating current @xmath33 remains due to surface conduction from @xmath15 to @xmath17 .
even if the time average @xmath34 vanishes , there will be time - dependent fluctuations @xmath35 with low - frequency noise power @xmath36 at low temperatures @xmath37 this is predominantly shot noise @xmath38 . in a scattering formulation the thermal and electrical transport properties can be expressed in terms of the matrix @xmath39 of transmission amplitudes from @xmath15 to @xmath17 , at energy @xmath40 relative to the fermi level .
the transmission matrix has a block structure in the electron - hole degree of freedom , @xmath41 the submatrix @xmath42 describes transmission of an electron as an electron , while @xmath43 describes transmission of an electron as a hole . at sufficiently small temperature and voltage the transmission matrix
may be evaluated at the fermi level ( @xmath44 ) and we have the landauer - type formulas @xcite @xmath45 eq . may equivalently be written in terms of the full transmission matrix , @xmath46
^ 2,\label{pshotdef2}\ ] ] with the help of the pauli matrix @xmath47 acting on the electron - hole degree of freedom . electron - hole symmetry at the fermi level equates @xmath48 with @xmath49 the pauli matrix that exchanges electrons and holes .
it follows that @xmath50 eqs . and
can therefore be combined into @xmath51 ^ 2 .
\end{split } \label{pg0ehsymm}\ ] ] since @xmath52 the term @xmath53 can be written as the trace of a positive definite matrix , @xmath54 so the dimensionless shot noise power @xmath55 is bounded from above by the dimensionless thermal conductance @xmath56 . in ref .
it was demonstrated that this inequality becomes a strict _ equality _ for a rank - one transmission matrix @xmath57 , as in 1d transmission via the unpaired majorana edge mode of a 2d topological superconductor.@xcite only particle - hole symmetry is needed in that case . in the next section
we will show that the combination of particle - hole symmetry and time - reversal symmetry achieves approximate equality on the 2d surface of a 3d topological superconductor , irrespective of the rank of @xmath57 .
the surface hamiltonian of majorana fermions in the @xmath58@xmath59 plane has the form@xcite @xmath60 with velocity @xmath61 , momentum operators @xmath62 , and pauli matrices @xmath63 acting on the spin degree of freedom .
( the @xmath64 unit matrix is @xmath65 and we have set @xmath66 to unity . ) we assume that there is no valley degeneracy of the surface states , so the surface spectrum consists of a single nondegenerate cone with dispersion relation @xmath67
. a disorder potential @xmath68 is forbidden by the combination of electron - hole symmetry and time - reversal symmetry , @xmath69 this insensitivity to impurity scattering is a unique property of a topological superconductor in symmetry class diii.@xcite a spatial modulation of the fermi velocity @xmath70 is allowed by symmetry,@xcite and this is the only source of scattering on the surface .
the normal metal has propagating modes labeled by a spin degree of freedom @xmath71 , electron - hole degree of freedom @xmath72 , and orbital degree of freedom @xmath73 .
the superconducting surface has only evanescent modes at the fermi level , because of the vanishing density of states . a majorana fermion with spin @xmath74 at point @xmath75
is coupled to the metal by the coupling matrix element @xmath76 .
the scattering matrix is given by@xcite @xmath77 near the fermi level where the energy dependence of the coupling matrix @xmath78 can be neglected .
the scattering matrix is unitary , @xmath79 , with electron - hole and time - reversal symmetries@xcite @xmath80 the corresponding symmetry relations for the coupling matrix are @xmath81 we now restrict ourselves to the fermi level , @xmath44 , and determine a condition on the coupling matrix @xmath78 that ensures the equality @xmath27 of the dimensionless shot noise power and thermal conductance . according to eqs . and
, a necessary and sufficient condition is that the electron - electron block @xmath82 of the transmission matrix product @xmath83 vanishes identically .
here we establish a _ sufficient _ condition involving only the coupling matrix : @xmath84 no requirements are made on the rank of @xmath57 , which may involve strong mode - mixing by the surface hamiltonian .
the combination of the electron - hole and time - reversal symmetries implies that [ calksymmetries ] @xmath85 so the matrix @xmath86 is antisymmetric and it anticommutes with @xmath87 .
because the only @xmath64 matrix with both these properties is identically zero , we conclude that @xmath88 vanishes if it is block - diagonal in @xmath64 matrices : @xmath89 the @xmath64 matrix @xmath90 acts on the spin degree of freedom of a majorana fermion at position @xmath75 on the superconducting surface
. we will refer to the condition as a _ locality condition _ on the coupling matrix product @xmath88 .
in the next section we will explicitly solve a simple model where the locality condition holds , but we argue that it is a natural assumption for a weakly disordered ns interface between the normal metal ( n ) and the topological superconductor ( s ) . if the disorder mean free path is large compared to the superconducting coherence length , a majorana fermion transmitted through the ns interface is locally converted into an electron - hole superposition via a @xmath91 matrix @xmath92 and then scattered nonlocally in the normal metal via a unitary matrix @xmath93 without mixing electrons and holes so @xmath93 commutes with @xmath94 and @xmath95 .
substitution of @xmath96 gives the desired locality to @xmath97 .
the transmission matrix @xmath57 refers to a 3d scattering problem , and this is how we will calculate it numerically later on . for an analytical treatment a reduction to an effectively 2d geometry is desirable . referring to fig .
[ fig_layout ] , in a typical thin - film geometry one has @xmath98 , so the contributions from the top and bottom surfaces in the @xmath58@xmath59 plane dominate over the contributions from the lateral surfaces in the @xmath99@xmath59 plane .
@xmath59 plane with tunnel coupling to a normal - metal in the region @xmath100 ( shaded , tunnel rate @xmath101 ) .
this 2d scattering geometry effectively describes the transmission from @xmath15 to @xmath17 via the top or bottom surface of the 3d topological superconductor in fig .
[ fig_layout ] .
the extension @xmath102 of the contact region is assumed to be sufficiently large that top and bottom surfaces can be treated independently .
( coupling of top and bottom surfaces is described in fig . [
fig_layout2dcoupled ] . ) ] the normal - metal contact region in the @xmath58@xmath99 plane is connected to the superconducting surface in the @xmath58@xmath59 plane at @xmath103 .
we ignore the curvature of the surface at this connection and replace the contact region by the region @xmath100 in the @xmath58@xmath59 plane .
tunneling into the metal electrode in the contact region is described by the effective hamiltonian @xmath104 with @xmath105 the unit step function . if the tunnel barrier is sufficiently transparent ( tunnel rate @xmath106 ) , a particle approaching @xmath103 via the top surface ( @xmath107 ) will enter the metal contact before reaching the bottom surface ( @xmath108 ) , so that we can treat top and bottom surfaces separately .
this produces the effectively 2d geometry of fig .
[ fig_layout2d ] , which we will analyze in the next subsection . the regime @xmath109 , when top and bottom surfaces can not be treated separately , is considered in sec .
[ sec_surfacecoupling ] .
the normal metal has a nonzero density of states at the fermi level , with a set of @xmath110 transverse momenta @xmath111 ( in the @xmath58-direction ) . at each @xmath111
there are four propagating modes , including the spin and electron - hole degree of freedom .
we collect the total number of @xmath112 mode indices in the label @xmath113 , with pauli matrices @xmath114 , @xmath115 acting , respectively , on the spin and electron - hole degree of freedom .
the scattering matrix element @xmath116 at energy @xmath40 relates an outgoing mode @xmath113 at @xmath59 to an incoming mode @xmath117 at @xmath118 .
the full @xmath119 scattering matrix @xmath120 describes both transmission ( when @xmath121 and @xmath122 or the other way around ) and reflection ( when @xmath123 or @xmath124 ) . in accord with eq .
it is given by @xmath125 in terms of a @xmath126 matrix green s function @xmath127 , @xmath128 and a @xmath129 coupling matrix @xmath130 .
the rank of the matrix @xmath131 is only half the rank of @xmath29 , because the majorana fermions on the superconducting surface lack the electron - hole degree of freedom of the dirac fermions in the normal metal.@xcite particle conservation ( unitarity ) requires that @xmath132 for @xmath133 , which is satisfied if @xmath134 although in the general treatment of the previous section we allowed for mode - mixing on the superconducting surface , for a tractable analytical calculation we now simplify to the hamiltonian with a uniform velocity @xmath61 . because other sources of scattering are excluded by the combination of electron - hole and time - reversal symmetry ,
the transverse momentum @xmath111 is not coupled by the effective hamiltonian and @xmath131 decomposes into @xmath64 @xmath111-dependent blocks @xmath135 .
this matrix green s function is calculated in app .
[ app_gsingle ] . to obtain the transmission matrix from @xmath15
to @xmath17 we must take @xmath121 and @xmath122 . at the fermi level
the result is [ gkappagamma ] @xmath136}{\xi\cosh lq+q\sinh lq } \nonumber\\ & \times\begin{pmatrix } -\xi-\kappa & iq\\ iq&\xi-\kappa \end{pmatrix},\label{gzeroresult}\\ \xi=\sqrt{q^2+\kappa^2},&\;\;\kappa=\tfrac{1}{2}\gamma / v,\;\;z > l/2,\;\;z'<-l/2.\label{xikappadef}\end{aligned}\ ] ] the @xmath119 transmission matrix @xmath137 at the fermi level follows from eq .
, @xmath138 with a @xmath111-dependent @xmath139 coupling matrix @xmath140 .
the unitarity constraint reads @xmath141 : same as fig .
[ fig_layout2d ] , but now with antiperiodic boundary conditions at @xmath142 to include the effect of a coupling of top and bottom surface via the contact region of finite length @xmath102 .
panel @xmath143 shows the correspondence between trajectories in the 2d and 3d representation .
coupling of the red and green trajectories , at the same transverse momentum @xmath111 , elevates the rank of the @xmath111-dependent transmission matrix from one to two . ]
the approach outlined above for the case of uncoupled top and bottom surfaces can be readily generalized to allow for a coupling of the two surfaces via the contact region . in the effective 2d representation the region of nonzero @xmath101 now extends over a finite interval , @xmath144.\label{gammazcoupled}\ ] ] a particle on the top surface crossing the contact region without being absorbed continues on the bottom surface .
the corresponding 2d geometry is shown in fig .
[ fig_layout2dcoupled ] .
it has a finite extent @xmath145 in the @xmath59-direction , with antiperiodic boundary conditions at @xmath146 to account for a @xmath147 berry phase .
the calculation of the green s function is given in app .
[ app_gdouble ] . instead of
we now have [ gkappagammacoupled ] @xmath148 the single - layer result is recovered in the limit @xmath149 .
the key distinction between the single - surface green s function and the coupled - surface result is that while both are @xmath64 matrices the latter is of rank two but the former is only of rank one ( one of the two eigenvalues of the matrix vanishes ) .
we use the 2d surface theory of the previous section to calculate the thermal conductance @xmath0 , including the effects of poor coupling to the metal contacts , strong coupling of top and bottom surfaces , and effects of a finite aspect ratio .
subject to the locality condition these results apply as well to the shot noise power @xmath150 .
the thermal conductance follows from the transmission matrix upon integration , @xmath151 because of the unitarity condition the coupling matrix drops out and only the @xmath64 matrix green s function enters .
substitution of the result for a single surface gives @xmath152 for @xmath153 the sum over transverse momenta may be replaced by an integration , @xmath154 , resulting in @xmath155 ) as a function of the coupling strength to the normal - metal contacts .
the solid curve is calculated from eq . , the dashed curve is the effective - length approximation . in the strong - coupling limit @xmath156 . ] the coupling strength of the superconductor to the metal contacts is quantified by the product @xmath157 . in the strong - coupling limit @xmath158 the thermal conductivity approaches the universal value @xmath159 this is the majorana fermion analogue@xcite of the dirac fermion conductivity in graphene.@xcite the effect of a finite coupling strength can be understood in terms of an effective length @xmath160 , so that @xmath161 see fig .
[ fig_gthermal ] .
we recover twice the single - surface plot in fig .
[ fig_gthermal ] .
the solid curves for finite @xmath102 are calculated from eq .
, the dashed curves are the approximation . ] if we allow for coupling of the top and bottom surfaces via the metal contact , we would use the green s function instead of eq .
, to arrive at @xmath162 there are now contributions from two surfaces in parallel , so @xmath163 in the large-@xmath164 limit . as shown in fig .
[ fig_gthermalcoupled ] , the finite-@xmath102 effect is accurately described by a reduction factor @xmath165 , @xmath166 deviations from the universal limit of the thermal conductivity appear even for strong coupling to the metal contacts , if the aspect ratio of the surface area is not large enough .
the relevant variable is the ratio @xmath167 of the perimeter @xmath168 of the metal contacts relative to their separation @xmath169 . of strong coupling to the metal contacts , as a function of the aspect ratio @xmath170 .
the curve is calculated from eq . .
a cube has @xmath171 and @xmath172 . ]
transverse momenta are quantized with antiperiodic boundary conditions , because of the @xmath147 berry phase , @xmath173 for strong coupling to the contacts ( @xmath174 ) we find from eq .
the thermal conductivity @xmath175 as the sum @xmath176}.\label{sigmathermalfiniter}\ ] ] as shown in fig .
[ fig_aspectratio ] , the universal limit is reached rather quickly ; for a cube geometry , @xmath177 , we are only 5% below the universal limit . in this model calculation
the general locality condition on the coupling matrix reads @xmath178 since @xmath179 upon fourier transformation .
the @xmath64 matrix @xmath180 , acting on the spin degree of freedom , may depend on @xmath59 but it should not depend on @xmath111 .
we only need to impose locality in @xmath58 , because in eq .
we have already taken a local coupling in @xmath59 .
the electron - hole and time - reversal symmetry constraints @xmath181 require that @xmath182 and this is only possible for a @xmath64 matrix@xcite if @xmath183 .
. gives @xmath184 and thus eq .
implies the equality @xmath185 all the results presented above for the thermal conductance then apply also to the shot noise power . within the 2d model calculation of this section
we can not ascertain that the locality condition holds . for that purpose
we need to perform a fully 3d calculation , as we will do in the next section .
our numerical simulation is based on the bogoliubov - de gennes hamiltonian@xcite @xmath186 discretized on a cubic lattice ( lattice constant @xmath187 , hopping energy @xmath188 ) .
the disorder potential is @xmath16 and the _ p_-wave pair potential is @xmath189 .
this is a generic model of a 3d topological superconductor in symmetry class diii , without spin - rotation symmetry but with electron - hole and time - reversal symmetries @xmath190
the geometry is that of fig .
[ fig_layout ] , in the normal - metal regions we set @xmath191 .
a tunnel barrier of height @xmath192 ( two lattice sites wide ) is introduced at the ns interfaces @xmath103 .
the scattering matrix is calculated using the kwant toolbox.@xcite we fixed the fermi energy at @xmath193 and took a relatively large pair potential @xmath194 to eliminate bulk conduction without requiring a large @xmath169 . at the ns interface
is the single fit parameter for the comparison .
these are calculations in the simplest case @xmath195 , @xmath196 , @xmath197 , @xmath198 . ] for a direct test of the analytical calculation from the previous section we first consider a translationally invariant system along the @xmath58-direction ( @xmath199 , no disorder ) .
results are shown in fig .
[ fig_widenumerics1 ] .
the analytical result eq
. describes the numerics very well , with the coupling strength @xmath164 as the single fit parameter .
this demonstrates the validity of the 2d representation of the 3d scattering problem , including the effect of coupling between top and bottom surfaces . , @xmath198 , @xmath195 , @xmath200 .
the magnetic field and spin - orbit coupling are varied in , respectively , the bottom and top panel .
this simulation demonstrates the inequality @xmath201 , with equality if either @xmath202 or @xmath203 vanishes . ]
we next investigate the extent to which the relation holds , still in the translationally invariant system , by adding to the hamiltonian the spin - orbit coupling @xmath204 in order to mix the modes from top and bottom surface .
( the same @xmath205 is added to superconducting and normal regions . ) note that @xmath205 preserves the electron - hole and time - reversal symmetries .
we break time - reversal symmetry by imposing on the normal metal the magnetic field @xmath206 , in the gauge @xmath207 . as shown in fig .
[ fig_widenumerics2 ] , both a nonzero @xmath203 and a nonzero @xmath202 are needed for a difference between dimensionless shot noise power and thermal conductance .
the nonzero @xmath203 is needed to couple the modes from top and bottom surface otherwise the transmission matrix would be of rank one and the equality would hold irrespective of whether time - reversal symmetry is broken or not.@xcite the nonzero @xmath202 is needed because of the argument from sec .
[ conditiononxi ] that mode coupling in the presence of time - reversal symmetry is not effective at violating the relation between shot noise power and thermal conductance .
, in a superconductor with @xmath208 ( @xmath196 , @xmath197 , @xmath200 ) . shot noise and thermal conductance differ by less than @xmath209 , even in the presence of significant disorder . ]
we now break translational invariance by adding a disorder potential @xmath16 , uniformly distributed in the interval @xmath210 , randomly fluctuating from site to site throughout the superconductor .
we also added disorder on the normal side of the ns interface ( in a sheet of width @xmath211 ) . because the calculations are now computationally more expensive we took a smaller superconductor , a cube of size @xmath212 .
results are shown in fig .
[ fig_cubenumerics ] . without disorder
the thermal conductivity is close to the limit expected from fig .
[ fig_gthermalcoupled ] for a cube aspect ratio : @xmath213 .
disorder has a significant effect , but the dimensionless shot noise and thermal conductance remain nearly indistinguishable .
as a particle that is its own antiparticle , a majorana fermion must be charge - neutral but it need not be in an eigenstate of charge .
this is a key distinction between a majorana fermion as a fundamental particle such as a neutrino , or as a composite quasiparticle in a superconductor . for the latter only the expectation value of the charge must vanish , so there may be quantum fluctuations of the charge . here
we have shown how we can exploit this property for a purely electrical detection of the majorana surface states in a 3d topological superconductor , obviating the difficulty of thermal measurements .
we like to think of the relation @xmath214 between thermal conductance and electrical shot noise power as the majorana counterpart of the electronic wiedemann - franz relation @xmath215 between thermal and electrical conductance.@xcite the analogy is quite direct : eq .
expresses the fact that a nonequilibrium electron transports energy and charge in a fixed ratio .
the same holds for eq . , with the electron charge @xmath216 replaced by the majorana charge variance @xmath217 .
there exists an altogether different `` wiedemann - franz type relation '' for majorana fermions , relating heat and _ particle _ currents rather than heat and _ charge _ currents : @xmath218 the `` particle conductance '' @xmath219 is not directly measurable ( since majorana fermions do not couple to the chemical potential ) , but it can be formally defined in terms of the landauer formula @xmath220 or in terms of an equivalent kubo formula.@xcite the factor @xmath12 is the `` topological '' or `` central '' charge @xmath221 of a majorana fermion.@xcite no such factor appears in eq . , because both the thermal conductance and the shot noise power are proportional to @xmath222 , so it drops out of the ratio .
one direction for future research is to generalize the relation to topological superconductors with more than a single species of majorana fermions on their surface .
this is a key difference with 3d topological insulators , which have a @xmath223 topological quantum number , so at most a single dirac cone on the surface .
in contrast , 3d topological superconductors have a @xmath224 topological quantum number , allowing for multiple majorana cones.@xcite another direction to explore is how the class - diii topological superconductors with majorana surface states considered here compare with the class - ci topological superconductors with dirac surface states . for eq .
the difference is simply a factor of two , to account for a central charge @xmath225 of dirac fermions.@xcite we do not expect such a simple correspondence for the relation . from the experimental point of view , the usefulness of eq .
is that it provides a purely electrical way to access the transport properties of majorana surface states .
the shot noise measurements should be performed at energies @xmath11 well below the superconducting gap @xmath189 . in cu@xmath226bi@xmath227se@xmath228
this is about 0.6 mev.@xcite shot noise dominates over thermal noise if @xmath229,@xcite so if one would perform the experiment at @xmath230 , a readily accessible temperature range @xmath231 would do .
this research was supported by the foundation for fundamental research on matter ( fom ) , the netherlands organization for scientific research ( nwo / ocw ) , and an erc synergy grant .
99 m. z. hasan and c. l. kane , _ topological insulators _ , rev .
phys . * 82 * , 3045 ( 2010 ) .
qi and s .- c .
zhang , _ topological insulators and superconductors _ , rev .
phys . * 83 * , 1057 ( 2011 ) .
m. sato and s. fujimoto , _ majorana fermions and topology in superconductors _ , j. phys .
. jpn . * 85 * , 072001 ( 2016 ) . c. w. j. beenakker and l. p. kouwenhoven , _ a road to reality with topological superconductors _ , nature phys .
* 12 * , 618 ( 2016 ) .
y. s. hor , a. j. williams , j. g. checkelsky , p. roushan , j. seo , q. xu , h. w. zandbergen , a. yazdani , n. p. ong , and r. j. cava , _ superconductivity in cu@xmath226bi@xmath227se@xmath228 and its implications for pairing in the undoped topological insulator _ , phys .
lett . * 104 * , 057001 ( 2010 ) .
s. sasaki , m. kriener , k. segawa , k. yada , y. tanaka , m. sato , and y. ando , _ topological superconductivity in cu@xmath226bi@xmath227se@xmath228 _ , phys .
* 107 * , 217001 ( 2011 ) .
n. levy , t. zhang , j. ha , f. sharifi , a. a. talin , y. kuk , and j. a. stroscio , _ experimental evidence for s - wave pairing symmetry in superconducting cu@xmath226bi@xmath227se@xmath228 single crystals using a scanning tunneling microscope _ , phys . rev
. lett . * 110 * , 117001 ( 2013 ) .
k. matano , m. kriener , k. segawa , y. ando , g .- q .
zheng , _ spin - rotation symmetry breaking in the superconducting state of cu@xmath226bi@xmath227se@xmath228 _ , nature phys .
( 2016 , in press ) .
t. asaba , b. j. lawson , c. tinsman , l. chen , p. corbae , g. li , y. qiu , y. s. hor , l. fu , and l. li , _ rotational symmetry breaking in a trigonal superconductor nb - doped bi@xmath227se@xmath228 _ , arxiv:1603.04040 .
a. r. akhmerov , j. p. dahlhaus , f. hassler , m. wimmer , and c. w. j. beenakker , _ quantized conductance at the majorana phase transition in a disordered superconducting wire _ ,
* 106 * , 057001 ( 2011 ) .
m. diez , i. c. fulga , d. i. pikulin , j. tworzydo , and c. w. j. beenakker , _ bimodal conductance distribution of kitaev edge modes in topological superconductors _ , new j. phys .
* 16 * , 063049 ( 2014 ) .
n. v. gnezdilov , b. van heck , m. diez , j. a. hutasoit , and c. w. j. beenakker , _ topologically protected charge transfer along the edge of a chiral p - wave superconductor _ , phys .
b * 92 * , 121406(r ) ( 2015 ) .
t. senthil and m. p. a. fisher , _ quasiparticle localization in superconductors with spin - orbit scattering _
b * 61 * , 9690 ( 2000 ) .
k. nomura , s. ryu , a. furusaki , and n. nagaosa , _ cross - correlated responses of topological superconductors and superfluids _ , phys .
lett . * 108 * , 026802 ( 2012 ) .
u. sivan and y. imry , _ multichannel landauer formula for thermoelectric transport with application to thermopower near the mobility edge _ , phys .
b * 33 * , 551 ( 1986 ) .
m. p. anantram and s. datta , _ current fluctuations in mesoscopic systems with andreev scattering _ , phys .
b * 53 * , 16390 ( 1996 ) . c. w. j. beenakker , _ random - matrix theory of majorana fermions and topological superconductors _ , rev
phys . * 87 * , 1037 ( 2015 ) .
g. e. volovik , _ the universe in a helium droplet _
( oxford university press , oxford , 2003 ) .
matthew s. foster , hong - yi xie , and yang - zhi chou , _ topological protection , disorder , and interactions : survival at the surface of three - dimensional topological superconductors _ ,
b * 89 * , 155140 ( 2014 ) .
r. queiroz and a. p. schnyder , _ helical majorana surface states of strongly disordered topological superconductors with time - reversal symmetry _ ,
b * 91 * , 014202 ( 2015 ) .
symmetry class diii refers to a superconductor with broken spin - rotation symmetry .
class - ci superconductors with preserved spin - rotation symmetry may also host gapless surface states , but these are dirac fermions rather than majorana fermions and do not have the same protection against disorder scattering .
r. nakai and k. nomura , _ disorder effects on thermal transport on the surface of topological superconductors by the self - consistent born approximation _ ,
b * 89 * , 064503 ( 2014 ) .
c. mahaux and h. a. weidenmller , _ shell - model approach to nuclear reactions _
( north - holland , amsterdam , 1969 ) .
the symmetry operations on the scattering matrix are taken in a time - reversally symmetric basis of the propagating modes in the normal metal .
if transverse momentum @xmath111 is used to label the modes , these operations should be accompanied by @xmath232 .
the factor of two difference between the number of independent modes in the normal metal and the superconductor is the reason that the mode - matching approach of ref . can not be applied here .
hong - yi xie , yang - zhi chou , and matthew s. foster , _
surface transport coefficients for three - dimensional topological superconductors _ , phys .
b * 91 * , 024203 ( 2015 ) . m. i. katsnelson , _ zitterbewegung , chirality , and minimal conductivity in graphene _ , eur
j. b * 51 * , 157 ( 2006 ) .
j. tworzydo , b. trauzettel , m. titov , a. rycerz , and c. w. j. beenakker , _ sub - poissonian shot noise in graphene _ ,
lett . * 96 * , 246802 ( 2006 ) .
any @xmath64 antisymmetric matrix @xmath180 is a scalar multiple of @xmath87 , so if we insist that @xmath180 anticommutes with @xmath87 we must have @xmath233 . c. w. groth , m. wimmer , a. r. akhmerov , and x. waintal , _ kwant : a software package for quantum transport _ , new j. phys . *
16 * , 063065 ( 2014 ) .
n. read and d. green , _ paired states of fermions in two dimensions with breaking of parity and time - reversal symmetries and the fractional quantum hall effect _ , phys .
b * 61 * , 10267 ( 2000 ) .
s. ryu , a. p. schnyder , a. furusaki , and a. w. w. ludwig , _ topological insulators and superconductors : ten - fold way and dimensional hierarchy _ , new j. phys .
* 12 * , 065010 ( 2010 ) .
m. titov , p. m. ostrovsky , i. v. gornyi , a. schuessler , and a. d. mirlin , _ charge transport in graphene with resonant scatterers _ , phys .
lett . * 104 * , 076802 ( 2010 ) .
we solve the differential equation @xmath234{\cal g}(z , z';q , e)\nonumber\\ = \openone \delta(z - z')\label{greenequation}\end{aligned}\ ] ] to obtain the @xmath64 matrix green s function that determines the transmission matrix of the majorana fermions between the normal - metal contacts .
( to simplify the notation we have set @xmath235 . ) for a similar calculation in graphene , see ref . .
we first consider in sec .
[ app_gsingle ] the case of uncoupled top and bottom surfaces , when the @xmath59-coordinate ranges over the entire real axis and the tunnel coupling in the contact region is given by @xmath236 in sec .
[ app_gdouble ] we incorporate the finite extension @xmath102 of the contact region , by setting @xmath144,\;\;|z|<l+d.\label{gammazcoupled2}\ ] ] antiperiodic boundary conditions at @xmath142 then couple the top and bottom surfaces .
@xmath239[m-\tfrac{1}{2}i\sigma_z+i\sigma_z\theta(z - z ' ) ] \nonumber\\ & \cdot{\cal p}_{0\leftrightarrow z'}\exp\left[\int_{0}^{z'}dz_2 \left ( -i\sigma_z\varepsilon(z_2)+q\sigma_y\right)\right],\label{ginfinity}\end{aligned}\ ] ] where @xmath240 indicates a monotically increasing or decreasing ordering of the @xmath59-dependent non - commuting operators , from @xmath241 leftmost to @xmath242 rightmost .
the matrix @xmath110 is determined by the requirement that @xmath243 : @xmath244(m\pm \tfrac{1}{2}i\sigma_z)=0.\label{mrequirement}\end{aligned}\ ] ] the row - spinor on the left - hand - side is orthogonal ( without taking complex conjugates ) to the column - spinor @xmath245 which is an eigenstate of @xmath246 the square root should be taken such that @xmath247 .
the result is @xmath248 with the definitions @xmath249 as a check , we take the limit @xmath250 , @xmath251 , when @xmath252 , as it should .
note the symmetry relations @xmath253 which ensure that the green s function satisfies the required time - reversal and electron - hole symmetries : @xmath254 to obtain the transmission matrix we set @xmath121 and @xmath122 : @xmath255 \exp\left[(l/2)\left ( ie\sigma_z+q\sigma_y\right)\right]\nonumber\\ & \cdot \left(m+ \tfrac{1}{2}i\sigma_z\right)\exp\left[(l/2)\left ( ie\sigma_z - q\sigma_y\right)\right]\exp\left[-(z'+l/2)\left ( ie\sigma_z-\tfrac{1}{2}\gamma\sigma_z - q\sigma_y\right)\right].\label{gsingleresultapp}\end{aligned}\ ] ] for @xmath44 this simplifies to eq . in the main text .
the green s function for coupled top and bottom surfaces is still of the form , with @xmath238 now given by eq . . instead of a decay at infinity
we now have the antiperiodic boundary conditions @xmath256 the condition on the matrix @xmath110 is replaced by @xmath257 \exp\left[d\left(i\varepsilon_0\sigma_z+ q \sigma_y\right)\right ] \exp\left[\tfrac{1}{2}l\left(ie\sigma_z+ q \sigma_y\right)\right ] \left ( m + \tfrac{1}{2}i \sigma_z\right ) \nonumber \\ & \quad = - \exp\left [ - \tfrac{1}{2}l\left(ie\sigma_z+q \sigma_y\right)\right ] \exp\left [ - d\left(i\varepsilon_0\sigma_z+ q \sigma_y\right)\right ] \exp\left [ - \tfrac{1}{2}l\left(ie\sigma_z+ q \sigma_y\right)\right ] \left ( m - \tfrac{1}{2}i \sigma_z\right ) , \label{pbce_m}\end{aligned}\ ] ] with solution @xmath258 for the transmission matrix we set @xmath259 , @xmath260 . the energy - dependent green s function
is then given by eq .
with @xmath110 from eq . .
at zero energy this produces the result from the main text .
|
we compare the thermal conductance @xmath0 ( at temperature @xmath1 ) and the electrical shot noise power @xmath2 ( at bias voltage @xmath3 ) of majorana fermions on the two - dimensional surface of a three - dimensional topological superconductor .
we present analytical and numerical calculations to demonstrate that , for a local coupling between the superconductor and metal contacts , @xmath4 ( with @xmath5 the lorenz number ) .
this relation is ensured by the combination of electron - hole and time - reversal symmetries , irrespective of the microscopics of the surface hamiltonian , and provides for a purely electrical way to detect the charge - neutral majorana surface states .
a surface of aspect ratio @xmath6 has the universal shot - noise power @xmath7 .
|
decays of @xmath1 mesons into @xmath20 and @xmath21 final states offer valuable information about strong interactions , weak interactions and possible new - physics ( np ) effects . in a series of recent papers @xcite
, we developed a strategy to address these aspects in a systematic manner .
it uses the following working hypotheses : * @xmath13 flavour symmetry of strong interactions ( but taking factorizable @xmath13-breaking corrections into account ) ; * neglect of penguin annihilation and exchange topologies .
we may gain confidence in these assumptions through internal consistency checks , which worked well within the experimental uncertainties for our previous numerical analyses .
since the @xmath1 factories reported updated results for several of the input quantities , we would like to explore the implications for the picture emerging from our strategy . for a detailed overview of the current experimental status of the @xmath0 observables ,
we refer the reader to the most recent compilation of the heavy flavour averaging group ( hfag ) @xcite
. we will give the updated numerical values for the quantities entering our strategy below . a somewhat surprising new development of this summer
is a new world average for @xmath22 , which went down by about 1@xmath23 thanks to an update by the belle collaboration @xcite , and is now given as follows : @xmath24 the corresponding straight line in the @xmath25@xmath26 plane of the generalized wolfenstein parameters @xcite is now on the lower side of the allowed region for the apex of the unitarity triangle ( ut ) of the cabibbo kobayashi
maskawa ( ckm ) matrix that follows from the usual indirect " fits @xcite . in view of this result
, we assume that this potential discrepancy is due to np in @xmath5@xmath6 mixing , and perform an analysis of the ut in section [ sec : gam - ut ] . to this end
, we use the data for the decays @xmath27 and @xmath28 , which allow us to determine the true " ut angle @xmath29 @xcite , serving as an input for our subsequent analysis . complementing this information with the measurement of @xmath4 through semi - leptonic @xmath1 decays
, we can construct the so - called reference unitarity triangle @xcite , and are in a position to convert the possibly emerging discrepancy for the ut into a cp - violating np physics phase in @xmath5@xmath6 mixing .
moreover , we may extract the true " values of @xmath30 and @xmath31 , where the latter serves as an input for our analysis of rare decays . in section [ sec : bpipi ]
, we then extract the hadronic parameters characterizing the @xmath32 system with the help of the @xmath33 isospin symmetry of strong interactions , and predict the cp - violating observables of the @xmath34 channel .
the results of this section are essentially theoretically clean , and serve as a testing ground for the calculation of the dynamics of the @xmath32 decays directly from qcd - related approaches , such as `` qcd factorization '' ( qcdf ) @xcite , the perturbative qcd approach ( pqcd ) @xcite , soft collinear effective theory " ( scet ) @xcite , or qcd sum rules @xcite .
the @xmath7 system and the status of the @xmath7 puzzle " are then the subject of section [ sec : bpik ] .
we find that our standard - model ( sm ) predictions for those decays that are only marginally affected by electroweak ( ew ) penguins are in accordance with the experimental picture , whereas this is not the case for the observables with prominent contributions from these topologies , in particular for the ratio @xmath8 of the cp - averaged rates of the neutral @xmath7 modes .
we show that this puzzle can still be resolved through np effects in the ew penguin sector with a large cp - violating new phase @xmath35 .
we have also a closer look at another hot topic the mixing - induced cp asymmetry of the @xmath36 channel and point out that this quantity depends strongly on the sign of the np phase @xmath35 . in particular , this asymmetry , which is found experimentally to be significantly smaller than the sm expectation , can be brought closer to the data by reversing the sign of @xmath35 .
moreover , we investigate whether significant non - factorizable @xmath13-breaking effects could have large impact on our analysis .
interestingly , we find that this is not the case , and note that such effects are also not indicted by the internal consistency checks of our working assumptions . in section [ sec : rare - bk - scenarios ] , we explore the interplay of the np in the ew penguin contributions to the @xmath7 system with rare @xmath1 and @xmath15 decays . to this end
, we apply the popular scenario that np enters the ew penguins through @xmath14-penguin topologies @xcite . in view of new experimental results
, we speculate on possible future scenarios . as in our previous analysis
, we find that @xmath37 , @xmath38 and @xmath39 are sensitive probes for these scenarios .
this is also the case for the mixing - induced @xmath36 cp asymmetry discussed in section [ sec : bpik ] .
finally , we summarize our conclusions in section [ sec : concl ] .
let us in view of the new result for @xmath22 first have a closer look at the ut .
the starting point of this analysis is the assumption that the possible discrepancy between ( [ s2b - exp ] ) and the ckm fits is due to cp - violating np contributions to @xmath5@xmath6 mixing ( although it is of course too early to say something definite on this issue at the moment ) @xcite .
for other recent analyses in this context , see @xcite .
we may then extract the general @xmath5@xmath6 mixing phase @xmath40 where @xmath41 could originate from physics beyond the sm , from the numerical value in ( [ s2b - exp ] ) , yielding @xmath42 .
here we have discarded a possible second solution around @xmath43 @xcite , which is disfavoured by recent @xmath1-factory data @xcite .
the phase ( [ phid - def ] ) enters the mixing - induced cp asymmetry of the @xmath44 channel , which arises in the following time - dependent rate asymmetry : @xmath45 in the sm , these observables can be written as ( for the explicit expressions , see @xcite ) @xmath46 where we have also given the most recent experimental numbers , and @xmath47 is a cp - conserving hadronic parameter , which measures sloppily speaking the ratio of the @xmath48 penguin to tree contributions .
let us now use the additional information which is provided by the @xmath49 decays .
the assumptions listed at the beginning of section [ sec : intro ] allow us then to derive @xmath50}_{\mbox{$7.5\pm 0.7 $ } } = \underbrace{-\frac{1}{\epsilon}\left[\frac{{\cal a}_{\rm cp}^{\rm dir}(b_d\to\pi^\mp k^\pm)}{{\cal a}_{\rm cp}^{\rm dir}(b_d\to\pi^+\pi^- ) } \right]}_{\mbox{$6.7\pm 2.0 $ } } \equiv h_{{\cal a}_{\rm cp}^{\rm dir}},\ ] ] where @xmath51 , and the ratio @xmath52 of the kaon and pion decay constants takes factorizable @xmath13-breaking corrections into account . in ( [ h - rel ] ) , we have indicated the numerical values following from the current data .
consequently , within the experimental uncertainties , this relation is also well satisfied by the new data , which gives us further confidence in our working assumptions .
the quantities @xmath53 and @xmath54 , which are fixed through the branching ratios and direct cp asymmetries , respectively , can be written as follows : @xmath55 if we complement this expression with ( [ cp - bpipi - dir - gen ] ) and ( [ cp - bpipi - mix - gen ] ) , we have sufficient information to determine @xmath29 , as well as @xmath56 @xcite . following these lines yields
@xmath57 @xmath58 consequently , @xmath53 and @xmath54 give results that are in good agreement with one another . as we discussed in @xcite , the solutions around @xmath59
can be excluded through an analysis of the whole @xmath0 system , which is also the case for the most recent data . in the following analysis , we will use @xmath60 corresponding to @xmath54 , as this is theoretically cleaner than the avenue offered by @xmath53 .
as we will see in subsection [ ssec : su3break ] , even large non - factorizable @xmath13-breaking corrections have a remarkably small impact on the numerical result in ( [ gamma - det ] ) .
the value for @xmath29 in ( [ gamma - det ] ) is somewhat larger than in @xcite , a significant part of the numerical shift can be explained by the new value for @xmath22 , as shown in fig .
[ fig : beta - gamma ] . before having a closer look at the whole set of hadronic parameters characterizing the @xmath32 system ,
let us first explore the implications of ( [ gamma - det ] ) for the apex of the ut in the @xmath25@xmath26 plane .
the interesting feature of this value of @xmath29 following from the cp asymmetries of the @xmath48 , @xmath49 system is that it does not receive in our scenario of np any significant np contributions .
consequently , it is the true " angle @xmath29 of the ut . in order to complete the determination of the `` true '' ut
, i.e. of the so - called reference ut @xcite , we use the ratio @xmath4 extracted from semi - leptonic tree - level @xmath1 decays .
although its values extracted from exclusive and inclusive decays are markedly different from each other , we use the following average @xcite : @xmath61 in the second column of table [ tab : rut ] , we list the values of @xmath25 , @xmath26 , and @xmath62 " for this value of @xmath4 and @xmath29 in ( [ gamma - det ] ) . for completeness , we also give the values of the lengths of the ut sides @xmath63 and @xmath64 and of the angle @xmath30 .
we observe that the value of @xmath62 that we obtain this way is significantly higher than the one in ( [ s2b - exp ] ) .
it corresponds to @xmath65 . in our scenario
, this difference is attributed to a non - vanishing value of the np phase @xmath41 in ( [ phid - def ] ) , where @xmath31 corresponds to @xmath62 . as seen in table [ tab : rut ] , our value of @xmath41 is compatible with the one found in @xcite , but our value of @xmath29 obtained here in a different manner is significantly higher .
an even larger value of @xmath29 following from @xmath66 , in the ballpark of @xmath67 , has been reported from an analysis using scet @xcite . in table
[ tab : rut ] , we also show the results for the reference unitarity triangle ( rut ) obtained in @xcite , where only cp violation in @xmath68 and @xmath4 in ( [ vub ] ) have been used as input .
the agreement between our analysis and the one in @xcite is almost too good . in obtaining the values in the column `` utfit rut '' we have used @xmath25 and @xmath26 from @xcite . in the last two columns in table [ tab : rut ]
we collect the results from @xcite within the sm and for the universal unitarity triangle ( uut ) in models with minimal flavour violation @xcite , where ( [ s2b - exp ] ) , @xmath4 and @xmath69 serve as inputs .
as already stated above , the values of @xmath29 are in both cases significantly smaller , while the values of @xmath30 are significantly larger than in the case of the rut .
the visibly increased value of @xmath64 relatively to the standard ut fits found by us in the case of the rut would require a small _ negative _ np contribution to the @xmath5@xmath6 mass difference @xmath70 and/or a slightly increased value of the non - perturbative parameter @xmath71 relevant for the ratio @xmath69 .
we look forward to improved data on the @xmath66 , @xmath72 system , @xmath4 , @xmath73 and @xmath74 in order to see whether the difference between the large value of @xmath29 found here and the one resulting from the uut and full ut fits could be interpreted as a clear signal of np . as determined in our strategy as a function of @xmath75 with all other experimental input parameters kept at their central values .
the lines correspond to the values in ( [ s2b - exp ] ) and ( [ gamma - det]).[fig : beta - gamma],width=302 ] .parameters of the reference ut ( rut ) determined through @xmath4 in ( [ vub ] ) and the cp asymmetries of the @xmath48 , @xmath49 system , yielding the value of @xmath29 in ( [ gamma - det ] ) , compared with the results of @xcite .
we show also the results of the full ut fit and of the universal unitarity triangle obtained in @xcite . [ cols="^,^,^,^,^,^",options="header " , ]
an attractive feature of the approach in @xcite is a direct connection between non - leptonic @xmath1 decays and rare @xmath1 and @xmath15 decays @xcite . assuming that the dominant np contributions enter through the @xmath14-penguin function @xmath76 , and using the renormalization - group evolution from scales @xmath77 to scales @xmath78
, we can directly investigate the impact of the modified ew penguin contributions in the @xmath7 modes on rare @xmath1 and @xmath15 decays .
proceeding in this manner we find that the value of @xmath79 in ( [ q - phi ] ) , which is preferred by the @xmath7 observables @xmath80 , requires the one - loop short - distance functions @xmath81 and @xmath82 to be at least as high as @xmath83 and @xmath84 in the sm .
the values in ( [ xy1 ] ) appear to violate the @xmath85 probability upper bounds @xmath86 obtained recently in the context of minimal flavour violation ( mfv ) @xcite . while our scenario of np having new complex phases goes beyond mfv , the inspection of the known formulae for @xmath87 shows that the upper bound on @xmath82 in ( [ xy2 ] ) is difficult to avoid
if the only np contribution resides in the ew penguins and the operator basis is the same as in the sm . for our analysis below
we will , therefore , use an only slightly more generous bound and impose @xmath88 .
taking then those values of @xmath79 from ( [ q - phi ] ) that also satisfy @xmath89 leaves us with @xmath90 note that this corresponds to a modest _ suppression _ of the magnitude of the ew penguin parameter relative to its new sm value of 0.58 .
another possible solution to the clash between ( [ xy1 ] ) and ( [ xy2 ] ) would be the introduction of new complex phases in the photon magnetic penguin contribution that has no impact on the @xmath91 decays but can influence @xmath87 .
this could weaken the tension between ( [ xy1 ] ) and ( [ xy2 ] ) subject to the bounds on the cp asymmetry in the @xmath92 decay , where the photon magnetic penguin plays an important rle @xcite .
another avenue one could explore would be the introduction of new operators in @xmath87 that would invalidate the bounds in ( [ xy2 ] ) .
for instance , new operators originating in higgs penguins in the mssm with a large @xmath93 could help here .
the impact of these new operators on @xmath87 turns out to be moderate when the constraints on their wilson coefficients from @xmath94 are taken into account @xcite .
still their presence can definitely weaken the bounds in ( [ xy2 ] ) , so that the values in ( [ xy1 ] ) are compatible with rare decay constraints in such a more complicated np scenario . in spite of these possibilities
, we will not explore them in the present paper because the predictive power of this more general np scenario is significantly smaller than of our scenario , unless a specific model is considered .
instead we will investigate how various modifications of @xmath79 , which allow us to satisfy the bounds in ( [ xy2 ] ) , influence our results for the observables of the @xmath91 system presented in section [ sec : bpik ] , and the predictions for rare decays discussed in detail in @xcite .
for this purpose , we have introduced three scenarios that represent possible future measurements of @xmath8 and @xmath95 : * scenario a : @xmath96 , @xmath97 , which is compatible with the present @xmath72 data and the rare decay bounds ( see ( [ q - phi - rd ] ) ) .
* scenario b : we assume that @xmath8 goes up , and take @xmath98 @xmath99 , which leads to @xmath100 , @xmath101 and some interesting effects in rare decays , as we shall see below .
this would , for example , occur if radiative corrections to the @xmath102 branching ratio enhance @xmath8 @xcite , though this alone would probably account for only about @xmath103 .
* scenario c : assume that both @xmath8 and @xmath95 move towards 1 ; taking @xmath104 leads to @xmath105 , @xmath106 .
the _ positive _ sign of the phase in this scenario distinguishes it strongly from both others .
the result of this exercise is contained in tables [ scentab1 ] and [ scentab2 ] : in table [ scentab1 ] , we show the values of a number of observables of the @xmath7 system in the three scenarios , while in table [ scentab2 ] , we show the corresponding values of the most interesting branching ratios for rare @xmath15 and @xmath1 decays . to this end
, we have used for the angle @xmath31 the value of @xmath62 in table [ tab : rut ] .
we observe that , in particular , the interplay of the @xmath107 modes is a very good and clean indication of which kind of np scenario to look for . due to the interference of charm and top contributions in @xmath108 , it is also the decay that can most naturally be suppressed ( though this is in contrast to the present experimental value ) . on the other hand , @xmath109 is always enhanced due to the large values of @xmath35 and the absence of the charm contribution .
concerning the observables of the @xmath72 system , @xmath110 offers a particularly interesting probe .
this cp asymmetry comes out very large in scenarios a and b , where the np phase is negative . on the other hand , the positive sign in scenario c brings this value closer to the data , in accordance with the features pointed out in subsection [ ssec : bpik - cp ] .
similarly the experimental value of @xmath111 favours a positive value of @xmath35 .
@xmath8 & 1.12 & @xmath112 & 1.03 & 1 & @xmath113 + @xmath95 & 1.15 & @xmath114 & 1.13 & 1 & @xmath115 + @xmath116 & 0.04 & @xmath117 ' '' '' & 0.06 & 0.02 & @xmath118 + @xmath119 & 0.06 & @xmath120 ' '' '' & 0.03 & 0.09 & @xmath121 + @xmath110 & @xmath122 & @xmath123 ' '' '' & @xmath124 & @xmath125 & @xmath126 + @xmath127 & 0.13 & 0.21 & 0.22 & 0.01 & @xmath128 + @xmath129 & @xmath130 & @xmath131 & @xmath132 & @xmath133 & @xmath134 + @xmath135 & @xmath136 & @xmath137 & @xmath138 & @xmath139 & @xmath140 ' '' '' + @xmath141 & @xmath142 & @xmath143 & @xmath144 & @xmath145 & @xmath146 + @xmath147 & @xmath148 & @xmath149 & @xmath150 & @xmath151 & @xmath152 + @xmath153 & @xmath154 & @xmath155 & @xmath156 & @xmath157 & @xmath158 + @xmath159 & @xmath160 & @xmath161 & @xmath162 & @xmath163 & @xmath164 ' '' '' + @xmath165 & @xmath166 & @xmath167 & @xmath168 & @xmath169 & @xmath170 + in table [ scentab2 ] , we show the effect of the various scenarios on selected rare decays .
our sm results differ slightly from the standard values because we use the ckm input from the first column in table [ tab : rut ] .
larger values of @xmath64 and @xmath26 than found in @xcite result in higher values of @xmath171 and @xmath109 , respectively , than found in @xcite and @xcite . the great sensitivity of the rare decay branching ratios to the parameters @xmath79 demonstrates clearly the impressive power of rare @xmath15 and @xmath1 decays to search for np effects in the ew penguin sector .
in this paper , we have reconsidered our analysis of @xmath0 and rare @xmath1 and @xmath15 decays in view of the new @xmath1-factory data for @xmath172 and the two - body modes as well as more stringent bounds on rare decays .
the main new messages from our analysis are as follows : * the @xmath44 and @xmath49 modes , which are marginally affected by ew penguins , allow us to determine @xmath3 . complementing this value with @xmath4
, we can determine the true unitarity triangle , allowing us to search for np contributions to @xmath5@xmath6 mixing ; we find a np phase @xmath173 .
* the @xmath7 puzzle , which is in particular reflected by the low experimental value of the ratio @xmath8 of the neutral @xmath7 branching ratios , persists
. it still points to np in the ew penguin sector , and favours a large np phase @xmath174 , although now also a value around @xmath175 can bring us rather close to the current experimental ranges of @xmath80 .
* @xmath10 would allow us to accommodate also the pattern of @xmath176 , which may be indicated by the @xmath1-factory data and received recently a lot of attention , although the measurements suffer still from large uncertainties .
a similar comment applies to the difference @xmath129 in ( [ deltaa ] ) of the direct cp asymmetries of the @xmath49 and @xmath12 modes , which could be increased for @xmath10 .
however , the latter cp asymmetry suffers from large uncertainties in our approach and does therefore not ( yet ) allow a stringent test . *
the internal consistency checks of the working assumptions of our strategy are satisfied at the level of @xmath177 , and can be improved through better data .
we studied the sensitivity of our numerical predictions of the @xmath7 observables on non - factorizable @xmath13-breaking effects of this order of magnitude , and found that the impact is surprisingly small .
consequently , it is in fact very exciting to speculate on np effects in the ew penguin contributions to the @xmath7 decays .
* in view of the fact that the parameters @xmath79 needed for the explanation of the low value of @xmath8 appear to imply rare decay branching ratios that violate the experimental bound from @xmath87 , we have explored various scenarios for @xmath79 that allow us to satisfy the rare decay constraints , but still give interesting results for both the @xmath91 decays and the rare @xmath15 and @xmath1 decays .
needless to say , our analysis of the @xmath32 modes and the determination of the angle @xmath29 described above are not affected by these modifications . on the other hand
, we find again that reversing the sign of the np phase @xmath35 brings the mixing - induced asymmetry of @xmath178 closer to the data .
our analysis demonstrates that the simultaneous study of non - leptonic @xmath1-decay branching ratios , the corresponding direct and mixing - induced cp asymmetries and rare @xmath15 and @xmath1 decays within a consistent phenomenological framework developed in @xcite , can with improved data shed light on new physics and the structure of qcd dynamics in non - leptonic @xmath1 decays . *
acknowledgements * + this work has been supported in part by the bundesministerium fr bildung und forschung under the contract 05ht4woa/3 and by the german - israeli foundation under the contract g-698 - 22.7/2002 .
buras , r. fleischer , s. recksiegel and f. schwab , _ phys .
lett . _ * 92 * ( 2004 ) 101804 ; _ nucl
_ * b697 * ( 2004 ) 133 . a.j .
buras , r. fleischer , s. recksiegel and f. schwab , _ acta phys .
_ * b36 * ( 2005 ) 2015 .
heavy flavour averaging group : http://www.slac.stanford.edu / xorg / hfag/. k. abe _ et al . _ [ belle collaboration ] , hep - ex/0507037 .
l. wolfenstein , _ phys .
lett . _ * 51 * ( 1983 ) 1945
buras , m.e .
lautenbacher and g. ostermaier , _ phys .
_ * 50 * ( 1994 ) 3433 .
et al . _
[ the _ utfit _ collaboration ] , _ jhep _ * 0507 * ( 2005 ) 028 ; for continuously updated analyses , see http://utfit.roma1.infn.it/. r. fleischer , _ phys .
_ * b459 * ( 1999 ) 306 ; _ eur .
j. _ * c16 * ( 2000 ) 87 .
r. fleischer and j. matias , _ phys .
_ * d66 * ( 2002 ) 054009 .
y. grossman , y. nir and m.p .
worah , _ phys .
_ * b407 * ( 1997 ) 307 .
t. goto , n. kitazawa , y. okada and m. tanaka , _ phys .
_ * d53 * ( 1996 ) 6662 ; + a.g .
cohen , d.b .
kaplan , f. lepeintre and a.e .
nelson , _ phys .
lett . _ * 78 * ( 1997 ) 2300 ; g. barenboim , g. eyal and y. nir , _ phys .
_ * 83 * ( 1999 ) 4486 .
m. beneke , g. buchalla , m. neubert and c.t .
sachrajda , _ phys .
_ * 83 * ( 1999 ) 1914 ; m. beneke and m. neubert , _ nucl .
_ * b675 * ( 2003 ) 333 .
li and h.l .
yu , _ phys .
_ * d53 * ( 1996 ) 2480 ; y.y .
keum , h .-
li and a.i .
sanda , _ phys .
_ * b504 * ( 2001 ) 6 ; y.y .
keum and h .- n . li , _ phys .
_ * d63 * ( 2001 ) 074006 .
bauer , d. pirjol and i.w .
stewart , _ phys .
_ * d65 * ( 2002 ) 054022 ; c.w .
bauer , s. fleming , d. pirjol , i.z .
rothstein and i.w .
stewart , _ phys .
_ * d66 * ( 2002 ) 014017 ; t. feldmann and t. hurth , _ jhep _ * 0411 * ( 2004 ) 037 .
a. khodjamirian , _ nucl .
_ * b605 * ( 2001 ) 558 ; a. khodjamirian , t. mannel and b. melic , _ phys .
_ * b571 * ( 2003 ) 75 .
g. colangelo and g. isidori , _ jhep _ * 9809 * ( 1998 ) 009 .
buras and l. silvestrini , _ nucl .
phys . _ * b546 * ( 1999 ) 299 .
buras , a. romanino and l. silvestrini , _ nucl .
_ * b520 * ( 1998 ) 3 .
buras , g. colangelo , g. isidori , a. romanino and l. silvestrini , _ nucl .
_ * b566 * ( 2000 ) 3 .
g. buchalla , g. hiller and g. isidori , _ phys .
_ * d63 * ( 2001 ) 014015 ; + d. atwood and g. hiller , lmu-09 - 03 [ hep - ph/0307251 ] .
r. fleischer and t. mannel , _ phys .
_ * b506 * ( 2001 ) 311 .
r. fleischer , g. isidori and j. matias , _ jhep _ * 0305 * ( 2003 ) 053 . k. agashe , m. papucci , g. perez and d. pirjol , hep - ph/0509117 .
bauer , i.z .
rothstein and i.w .
stewart , _ phys .
lett . _ * 94 * ( 2005 ) 231802 ; hep - ph/0510241 .
a. j. buras , p. gambino , m. gorbahn , s. jager and l. silvestrini , _ phys .
_ * b500 * ( 2001 ) 161 .
g. dambrosio , g. f. giudice , g. isidori and a. strumia , _ nucl .
_ * b645 * ( 2002 ) 155 .
buras and r. fleischer , _ eur .
j. _ * c11 * ( 1999 ) 93 .
m. gronau , d. pirjol and t.m .
yan , _ phys .
_ * d60 * ( 1999 ) 034021 [ erratum - ibid .
* d69 * ( 2004 ) 119901 ] .
r. fleischer , _ phys .
_ * b365 * ( 1996 ) 399 .
a. ali , e. lunghi and a.y .
parkhomenko , _ eur .
j. _ * c36 * ( 2004 ) 183 .
chiang , m. gronau , j.l .
rosner and d.a .
suprun , _ phys .
_ * d70 * ( 2004 ) 034020 .
y. l. wu and y. f. zhou , _ phys .
_ * d72 * ( 2005 ) 034037 ; _ phys .
_ * d71 * ( 2005 ) 021701 .
r. fleischer and t. mannel , _ phys .
_ * d57 * ( 1998 ) 2752 .
r. fleischer , _ eur .
j. _ * c6 * ( 1999 ) 451 .
m. neubert , _ jhep _ * 9902 * ( 1999 ) 014 .
r. fleischer and s. recksiegel , _ eur .
j. _ * c38 * ( 2004 ) 251 ; _ phys .
_ * d71 * ( 2005 ) 051501 .
buras and r. fleischer , _ eur .
j. _ * c16 * ( 2000 ) 97 .
m. neubert and j.l .
rosner , _ phys .
_ * b441 * ( 1998 ) 403 ; _ phys .
lett . _ * 81 * ( 1998 ) 5076 .
m. beneke , _ phys .
_ * b620 * ( 2005 ) 143 .
m. gronau , y. grossman and j.l .
rosner , _ phys .
_ * b579 * ( 2004 ) 331 .
p. ball and r. zwicky , _ phys .
_ * d71 * ( 2005 ) 014015 .
buras , r. fleischer , s. recksiegel and f. schwab , _ eur .
j. _ * c32 * ( 2003 ) 45 . c. bobeth , m. bona , a.j .
buras , t. ewerth , m. pierini , l. silvestrini and a. weiler , _ nucl .
_ * b726 * ( 2005 ) 252 .
kagan and m. neubert , _ phys .
_ * d58 * ( 1998 ) 094012 .
g. hiller and f. krger , _ phys .
_ * d69 * ( 2004 ) 074020 .
e. baracchini and g. isidori , hep - ph/0508071 .
buras , m. gorbahn , u. haisch and u. nierste , hep - ph/0508165 .
|
we analyse the @xmath0 modes in the light of the most recent @xmath1-factory data , and obtain the following new results : ( i ) the @xmath2 modes prefer @xmath3 , which together with @xmath4 allows us to determine the `` true '' unitarity triangle and to search for cp - violating new - physics contributions to @xmath5@xmath6 mixing ; ( ii ) the @xmath7 puzzle reflected in particular by the low experimental value of the ratio @xmath8 of the neutral @xmath7 rates persists and still favours new physics in the electroweak penguin sector with a new cp - violating phase @xmath9 , although now also @xmath10 can bring us rather close to the data ; ( iii ) the mixing - induced @xmath11 cp asymmetry is a sensitive probe of the sign of this phase , and would currently favour @xmath10 , as well as the direct cp asymmetry of @xmath12 , which suffers , however , from large hadronic uncertainties ; ( iv ) we investigate the sensitivity of our @xmath7 analysis to large non - factorizable @xmath13-breaking effects and find that their impact is surprisingly small so that it is indeed exciting to speculate on new physics ; ( v ) assuming that new physics enters through @xmath14 penguins , we study the interplay between @xmath7 and rare @xmath1 , @xmath15 decays and point out that the most recent @xmath1-factory constraints for the latter have interesting implications , bringing us to a few scenarios for the future evolution of the data , where also the mixing - induced cp violation in @xmath11 plays a prominent rle .
cern - ph - th/2005 - 241 + tum - hep-610/05 + mpp-2005 - 156 + hep - ph/0512032 * andrzej j. buras,@xmath16 robert fleischer,@xmath17 stefan recksiegel@xmath16 and felix schwab@xmath18 * @xmath16 _ physik department , technische universitt mnchen , d-85748 garching , germany _
@xmath17 _ theory division , department of physics , cern , ch-1211 geneva 23 , switzerland _
@xmath19 _ max - planck - institut fr physik werner - heisenberg - institut , d-80805 munich , germany
_ december 2005
|
_ preference aggregation theory _ is concerned with aggregating individual preferences into a ( collective ) social preference , which is then maximized to yield a set of best alternatives .
the theory investigates the extent to which social preferences inherit desirable properties from individual preferences . we typically restrict ( strict ) individual and social preferences to those asymmetric relations @xmath0 on a set @xmath1 of alternatives that are either ( i ) acyclic or ( ii ) transitive or ( iii ) negatively transitive . by @xmath2 .
@xmath0 is asymmetric iff @xmath3 is complete ( reflexive and total ) .
( i ) @xmath0 is _ acyclic _ if for any finite set @xmath4 , whenever @xmath5 , , @xmath6
, we have @xmath7 . if @xmath0 is acyclic , it is asymmetric and irreflexive .
( ii ) @xmath0 is _ transitive _ if @xmath8 and @xmath9 imply @xmath10 .
when @xmath0 is transitive , we say @xmath3 is _ quasi - transitive_. ( iii ) @xmath0 is _ negatively transitive _ if @xmath11 and @xmath12 imply @xmath13 .
@xmath0 is negatively transitive iff @xmath3 is transitive .
] of the properties ( i ) , ( ii ) , and ( iii ) for asymmetric preferences , _ negative transitivity _ is the most demanding .
arrow s theorem ( @xcite ) points out the difficulty of aggregating preferences for more than two alternatives while preserving asymmetry and negative transitivity .
_ acyclicity _ is the least demanding of the properties ; it is necessary and sufficient for the existence of a maximal element on _ every finite subset _ of alternatives .
nakamura number _ plays a critical role in the study of preference aggregation rules with _
acyclic _ social preferences .
consider a _ simple game _ ( voting game ) @xmath14 , a collection of `` winning '' coalitions in a set @xmath15 of players .
combining the game with a set @xmath1 of alternatives and a profile @xmath16 of individual preferences , one obtains a _ simple game with preferences _
@xmath17 , for which one can define the social preference @xmath18 ( dominance relation ) and the _ core _
@xmath19 ( the set of maximal alternatives with respect to @xmath18 ) .
nakamura s theorem ( @xcite ) gives a restriction on the number of alternatives that the set of players can deal with rationally ( theorem [ nakamura - thm ] ) : the core of a simple game with preferences is always ( i.e. , for all profiles of acyclic preferences ) nonempty if and only if the number of alternatives is finite and below a certain number , called the _ nakamura number _ of the simple game .
the theorem thus gives a condition ( corollary [ nakamura - cor ] ) for the social preferences to inherit acyclicity from individual preferences . to deal with an empty core ( or cycles in social preferences )
, several authors have investigated solutions different from the core .
we pick two examples for which there have been recent developments .
first , @xcite proposes a procedure in which one deletes particular instances of preferences until the resulting subrelation is acyclic ( alternatively , transitive or negatively transitive ) , and collects the maximal elements of all such ( maximal acyclic ) subrelations .
second , taking voters foresight into consideration , @xcite proposes the _ stability set _ , a superset of the core .
all these investigations focus on treating cyclic _ social _ preferences , assuming that _ individual _ preferences satisfy a rationality property at least as strong as acyclicity . in this paper , we propose a _ preference aggregation theory without acyclicity_. in contrast to the authors cited in the preceding paragraph , we do not attempt to remove cycles in ( social and even individual ) preferences . we retain the usual framework ( e.g. , * ? ? ?
* section ii.2 ) in which a set @xmath1 of underlying alternatives is distinguished from an _ agenda _ ( opportunity set ) @xmath20 with which a group @xmath15 of players are confronted .
in particular , fixing a simple game @xmath14 and a set @xmath1 , we focus on the aggregation methods that assign the core @xmath21 or the _ core without majority dissatisfaction _ @xmath22 ( introduced later ) . ] to each @xmath23 satisfying a certain assumption .
the assumption , which is actually a condition concerning a pair @xmath24 , is that _ every player s preference @xmath25 has a maximal element in @xmath26 _ : the _ maximal set _ @xmath27 ( the set of maximal elements of the preference ) is nonempty for each @xmath28 .. when @xmath1 is infinite , there is an acyclic preference that has no maximal element on @xmath1 .
similarly , the property of having a maximal element on @xmath26 is independent of the property of having a maximal element on @xmath1 . ] this is a rather direct assumption at the individual level , which is to be inherited to the requirement at the social level that there is a chosen alternative for the given pair .
the assumption is distinctive in that it involves an agenda @xmath26 as well as a profile @xmath29 .
for this reason , it does not fit the following _ standard scenario _ in social choice theory @xcite very well : before knowing an agenda , the players discover and report their own preferences ( on @xmath1 ) to the planner ; having obtained a choice rule that assigns a nonempty subset to every agenda , the planner applies the rule to a particular agenda @xmath26 . since the planner does not generally know whether a pair @xmath24 satisfies the assumption until she faces the agenda @xmath26 , what she obtains immediately after learning the profile @xmath29 of preferences is only a _ partial _ choice rule , which might assign an empty set to some agenda .
an _ alternative scenario _ that the assumption fits well is the following : the planner first presents an agenda @xmath26 to the players , who then discover and report their own preferences ( for alternatives in @xmath26 at least ) to the planner ; the planner then makes a choice .
this is perhaps a closer description of actual collective decisions .
while failing to produce a choice rule at an intermediate stage , the scenario has some advantages over the standard scenario .
first , the alternative scenario can deal with context - dependent choices based on multiple _ rationales _ ( preferences belonging to the same individual ) more easily , where the context is given by an agenda @xcite .
the problem with the standard scenario is that a player is supposed to report a single preference for the whole set @xmath1 , when she might actually have an irreducible set of multiple rationales .
second , as @xcite writes , `` ideally , one could observe all preferences among the available alternatives , but there would be no way to observe preferences among alternatives not feasible for society , '' even if each player has a single preference .
this argument justifies the alternative scenario either directly or via _ arrow s iia _ ( independence of irrelevant alternatives ) , which requires that social choice from an agenda depends only on the individual preferences restricted to the agenda .
both the core and the core without majority dissatisfaction satisfy arrow s iia . for all @xmath28 , we have @xmath30 and @xmath31 .
the assertion for @xmath32 is a corollary of lemma [ c+maximal ] .
the reader should not confuse arrow s iia with the iia ( sometimes called _ property @xmath33 _ ) for choice rules , discussed by @xcite , for example . ] or on @xmath26 .
if a preference were defined on an agenda , however , a more straightforward formulation would be to remove the symbol @xmath26 from the framework and call that agenda @xmath1 .
doing so , however , would make the connection between the alternative scenario and the framework much less clear .
for this reason , we _ formally _ define preferences on @xmath1 instead of on @xmath26 . ]
the _ core without majority dissatisfaction _
( definition [ c+ ] ) is obtained by deleting from an agenda all the alternatives that are non - maximal for all _ individual _ players in a large ( winning ) coalition .
it is ( lemma [ c+in_c ] ) a strengthening or subset of the _ core _ , obtained by deleting from the agenda all the alternatives that are non - maximal for a large ( winning ) coalition of players _
collectively_. consequently , it only chooses pareto efficient alternatives from an agenda , unlike other solutions such as the stability set @xcite .
the core without majority dissatisfaction is a simple solution concept that treats the maximal sets @xmath34 of the players in a better - behaved way than the core does .
( it is reasonable to pay attention to such sets , since they are the very objects that we assume to be nonempty . ) first , unlike the core , the core without majority dissatisfaction depends only on the players maximal sets ( lemma [ c+maximal ] ) .
second , each alternative in the core without majority dissatisfaction belongs to someone s maximal set ( lemma [ c+in_maximals ] ) .
the same is not true for the core ( examples [ ex : core1 ] and [ ex : core2 ] ) .
the main results of the paper are similar to nakamura s theorem ( @xcite ) , except that they consider _
profiles for _ an agenda @xmath26profiles of ( not necessarily acyclic ) preferences that have maximal elements on the agenda .
_ the first main result _ , theorem [ nakamura - max ] , is about the original nakamura number for simple games @xmath14 defined on an algebra of coalitions .
it asserts that the following statements are equivalent : ( i ) the number of alternatives in the agenda @xmath26 is less than the nakamura number @xmath35 ; ( ii ) the core @xmath22 without majority dissatisfaction is nonempty for all profiles @xmath29 for the agenda ; ( iii ) the core @xmath21 is nonempty for all profiles @xmath29 for the agenda .
regardless of which choice rule is used , the nakamura number therefore measures the extent ( the size of an agenda ) to which simple games will assuredly choose some alternative from the agenda , whether individual preferences are assumed to be acyclic or to have maximal elements .
theorem [ nakamura - max ] is remarkable for the following reasons : first , it demonstrates that one can obtain a significant result without assuming acyclic preferences .
though neglected in the literature , a preference aggregation theory without acyclicity has potential .
second , the _ general _ nonemptyness of the core implies the _ general _ nonemptyness of the _ strengthening _ of the core .
the core without majority dissatisfaction is as satisfactory as the core according to this criterion .
third , restricting preferences to those with maximal elements allows us to drop the awkward condition that the agenda is finite .
unlike theorem [ nakamura - thm ] , theorem [ nakamura - max ] gives a condition for the general nonemptyness of the core for _ infinite _ , as well as finite , agenda .
fourth , it fits the alternative scenario .
it gives a condition for the planner to be assured of the existence of a chosen alternative as soon as she presents an agenda to the players ( i.e. , before she learns their preferences , supposing that they have maximal elements ) .
fifth , our framework is very general .
like @xcite , we impose no conditions such as monotonicity or properness on simple games .
unlike @xcite , we consider arbitrary sets of players and _ arbitrary algebras _ of coalitions . _
the second main result _
, theorem [ nakamura - max - extended ] , is about the _ kappa number _ ( definition [ kappa ] ) , an extension of the nakamura number to the even more general framework that distinguishes the collection @xmath36 of the sets of players for which one can assign winning / losing status from the algebra @xmath37 of ( identifiable ) coalitions . the kappa number @xmath38 is defined for a collection @xmath39 of winning sets .
the result asserts that the following two statements are equivalent : ( i ) the number of alternatives in the agenda @xmath26 is less than the kappa number @xmath38 ; ( ii ) the core @xmath40 without majority dissatisfaction is nonempty for all profiles @xmath29 for the agenda .
it also asserts that the above two statements imply , but are not implied by : ( iii ) the core @xmath41 is nonempty for all profiles @xmath29 for the agenda .
theorem [ nakamura - max - extended ] gives another sense in which the core without majority dissatisfaction behaves better than the core .
computing the kappa number is not an easy task in general .
so , in applying the theorem , lemma [ nu - kappa - nu ] is useful ; it estimates the kappa number from above and below in terms of the nakamura numbers .
the notion of _ ordinal numbers _ ( or _ ordinals _ ) generalizes that of natural numbers .
this section presents some facts about ordinals .
consult a textbook for axiomatic set theory ( e.g. , @xcite ) for more systematic treatment .
we start by introducing the first `` few '' ordinals .
the natural numbers are constructed from @xmath42 as follows : @xmath43 , @xmath44 , @xmath45 , @xmath46 , @xmath47 , etc .
the first ordinal number that is not a natural number is the set @xmath48 of natural numbers .
we can continue the process to obtain @xmath49 , @xmath50 , etc .
we then have @xmath51 , @xmath52 , , @xmath53 , , @xmath54 . for a given ordinal @xmath33 , its _ successor _
@xmath55 always exists and is an ordinal . for a given ordinal @xmath33
, @xmath56 does not necessarily exist : if there is an ordinal @xmath57 such that @xmath58 , then @xmath33 is a _
successor ordinal _ ; otherwise , it is a _
limit ordinal_. every natural number except @xmath59 is a successor ordinal . both @xmath60 and @xmath61
are limit ordinals . but
@xmath62 , @xmath63 , etc . are successor ordinals
. define @xmath64 if and only if @xmath65 .
@xmath66 has all the properties of a linear order .
every set @xmath67 of ordinal numbers are _ well - ordered _ by @xmath66 , that is , every nonempty subset of @xmath67 has a @xmath66-least element .
each ordinal @xmath33 has the property that @xmath68 if @xmath33 is a successor ordinal , say @xmath69 , then as a set , it has a greatest element , namely @xmath57 .
if @xmath33 is a limit ordinal , then it does not have a greatest element , and @xmath70 . a function
whose domain is an ordinal @xmath33 is called a _ ( transfinite ) sequence _ of length @xmath33 .
two sets @xmath71 and @xmath72 are _ equipotent _ if there is a bijection ( one - to - one and onto function ) from @xmath71 to @xmath72 .
an ordinal number @xmath33 is an _ initial _
ordinal if it is not equipotent to any @xmath73 .
for example , @xmath60 is an initial ordinal , because it is not equipotent to any natural number . @xmath62
not initial , because it is equipotent to @xmath60 .
similarly , none of countable ordinals @xmath63 , @xmath74 , @xmath61 , @xmath75 , @xmath76 , is initial .
the _ cardinal number _ of a set @xmath71 , denoted @xmath77 , is the unique initial ordinal equipotent to @xmath71 .
in particular , if @xmath71 is countable , then @xmath78 .
there are arbitrarily large cardinal numbers .
infinite cardinal numbers form a transfinite sequence of _ alephs _ @xmath79 , with @xmath33 ranging over all ordinal numbers .
we have @xmath80 .
appendix [ card - lim ] gives a proof of the following : [ lem : card - lim ] an infinite cardinal number is a limit ordinal . without defining the _ ordinal sum _ and the _ cardinal sum _ here ,
let us just mention the following useful lemma ( proved in appendix [ sumsum ] ) : [ lem : sumsum ] @xmath81 , where the sum on the left side is the ordinal sum and the sum on the right is the cardinal sum .
let @xmath15 be an arbitrary nonempty set of players and @xmath82 an arbitrary boolean algebra of subsets of @xmath15 ( so @xmath37 includes @xmath15 and is closed under union , intersection , and complementation ) .
the elements of @xmath37 are called * coalitions*. intuitively , they are the observable or identifiable or describable subsets of players . a ( @xmath37)-*simple game * @xmath14 is a subcollection of @xmath37 such that @xmath83 and @xmath84 .
the elements of @xmath14 are said to be * winning * , and the other elements in @xmath37 are * losing*. a simple game @xmath14 is * weak * if the intersection @xmath85 of the winning coalitions is nonempty .
let @xmath1 be a ( finite or infinite ) set of * alternatives * , with cardinal number @xmath86 .
_ in this paper , a * ( strict ) preference * is an _ asymmetric _
relation @xmath0 on @xmath1 _ : if @xmath8 ( `` @xmath87 is preferred to @xmath88 '' ) , then @xmath89 .
a relation @xmath0 is * total * if @xmath90 implies @xmath8 or @xmath91 .
an asymmetric relation is a * linear order * if it is transitive and total .
a binary relation @xmath0 on @xmath1 is _ acyclic _ if for any finite set @xmath4 , whenever @xmath5 , , @xmath6
, we have @xmath7 .
acyclic relations are preferences since they are asymmetric ( and irreflexive ) .
let @xmath92 be the set of _ acyclic _ preferences on @xmath1 .
a * profile * is a list @xmath16 of individual preferences @xmath25 .
intuitively , @xmath93 means that player @xmath28 prefers @xmath87 to @xmath88 at profile @xmath29 .
a profile @xmath29 is * ( @xmath37)-measurable * if @xmath94 for all @xmath87 , @xmath95 .
denote by @xmath96 the set of all measurable profiles of _ acyclic _ preferences .
an * agenda * ( or `` budget set '' or `` opportunity set '' ) @xmath26 is a subset of @xmath1 .
note that a preference , when restricted to the elements in @xmath26 , defines an asymmetric relation on @xmath26 .
let @xmath20 be an agenda .
an alternative @xmath97 is said to be a * maximal * element of @xmath26 with respect to a binary relation @xmath0 ( written @xmath98 though @xmath26 is often dropped ) if there does not exist an alternative @xmath99 such that @xmath91 .
let @xmath100 be the set of * preferences for @xmath26 * , i.e. , _ asymmetric _ relations @xmath0 on @xmath1 _ that has a maximal element of @xmath26_. on @xmath1 , despite the fact that the set of maximal elements of @xmath26 with respect to @xmath0 depends only on the restricted relation @xmath101 .
] let @xmath102 be the set of * profiles for @xmath26 * , i.e. , measurable profiles @xmath103 of preferences @xmath25 for @xmath26 .
a * ( @xmath37)-simple game with ( ordinal ) preferences * is a list @xmath104 of a @xmath37-simple game @xmath105 , a subset @xmath26 of alternatives , and a profile @xmath16 .
given the simple game with preferences , we define the ( not necessarily asymmetric ) * dominance * relation @xmath18 on @xmath1 by @xmath106 if and only if there is a winning coalition @xmath107 such that @xmath108 for all @xmath109 .
need not be winning since we do not assume @xmath14 is monotonic .
@xcite study nakamura s theorem , adopting the notion of dominance that requires the above coalition to be winning .
their dominance relation distinguishes the game from its monotonic cover , while the classical dominance @xmath18 does not .
the two notions are equivalent if and only if the game is monotonic . ]
the * core * @xmath110 of the simple game with preferences is the set @xmath111 of undominated alternatives : @xmath112 an alternative @xmath97 is * pareto in @xmath26 * if there exists no @xmath99 such that @xmath113 .
it is easy to prove that any alternative in @xmath110 or in @xmath114 is pareto in @xmath26 .
a * ( preference ) aggregation rule * is a map @xmath115 from profiles @xmath29 of preferences to binary relations ( _ social preferences _ ) @xmath116 on the set @xmath1 of alternatives .
for example , the mapping @xmath117 from profiles @xmath118 of acyclic preferences to dominance relations @xmath18 is an aggregation rule .
@xcite gives a condition for a @xmath119-simple game with preferences to have a nonempty core for any profile @xmath29 of _ acyclic _ preferences . to state nakamura s theorem
, we define the * nakamura number * @xmath35 of a @xmath37-simple game @xmath14 to be the size of the smallest collection of winning coalitions having empty intersection - least element . ]
@xmath120 if @xmath121 ( i.e. , if @xmath14 is _ nonweak _ ) ; otherwise , set @xmath122 , which is understood to be greater than any cardinal number . by the assumption that @xmath83 and @xmath84 , we have @xmath123 .
it is easy to prove the following lemma : [ nakamura - bounds ] if @xmath14 is a nonweak @xmath37-simple game , then @xmath124 and @xmath125 .
the following theorem extends nakamura s result @xcite for @xmath126 : [ nakamura - thm ] let @xmath105 be a simple game and @xmath20 an agenda
. then the core @xmath110 is nonempty for all measurable profiles @xmath118 of _ acyclic _ preferences if and only if _ @xmath26 is finite _ and @xmath127 .
since @xmath18 is acyclic if and only if the set @xmath21 of maximal elements of @xmath26 with respect to @xmath18 is nonempty for every finite subset @xmath26 of @xmath1 , we have : [ nakamura - cor]the dominance relation @xmath18 is acyclic for all @xmath118 if and only if @xmath127 for all _ finite _ @xmath20 .
in this section , we first introduce the notion of the _ core without majority dissatisfaction _ , a strengthening of the core .
we then compare the two notions of the core , focusing on how they treat the maximal elements of individual preferences .
we consider @xmath37-simple games @xmath104 with preferences , given for each profile @xmath29 ( not necessarily in @xmath102 ) .
an alternative @xmath97 is _ not _ in the core @xmath110 if @xmath87 is not maximal with respect to the dominance relation @xmath18 : there are @xmath99 and a winning coalition @xmath107 such that for all @xmath109 , @xmath128 and @xmath129 .
( that is , @xmath97 is not in @xmath21 if for some @xmath99 , the set @xmath130 contains a winning coalition . )
so , an alternative @xmath87 ( e.g. , @xmath131 in examples [ ex : core1 ] and [ ex : core2 ] below ) can be in the core even if there is a winning coalition contained in the set of players @xmath28 that prefer some @xmath132 to @xmath87 , as long as @xmath132 is different among the players . to exclude such an @xmath87 from the core
, we modify the definition : [ c+ ] an alternative @xmath97 is in the * core @xmath22 without majority dissatisfaction * if there is no winning coalition @xmath107 such that for all @xmath109 , there exists some @xmath133 satisfying @xmath129 . , an alternative must be a maximal element for at least one individual in each @xmath107 .
so we can rewrite @xmath134 . ] in other words , @xmath97 is in @xmath22 if the set @xmath135 for some @xmath136 of players for whom @xmath87 is non - maximal ( players `` dissatisfied with @xmath87 '' ) contains no winning coalition.the core without majority dissatisfaction consists of those alternatives not rejected by the following scheme ( when the players behave sincerely ) : after proposing an agenda @xmath26 , for each alternative @xmath97 , the planner does the following : ( i ) she proposes @xmath87 to the players ; ( ii ) she asks each player @xmath28 whether @xmath28 is `` dissatisfied '' with @xmath87 ( which , by definition , means whether @xmath28 finds some alternative in @xmath26 better than @xmath87 ) , without asking what alternative @xmath132 is better ; ( iii ) if a winning coalition of players are `` dissatisfied '' with @xmath87 , then the planner rejects @xmath87 . under this scheme ,
the planner elicits individual preferences in a very incomplete manner ( as is usual with real - world decisions ) .
also , the members of the winning coalition are only united in their opposition to @xmath87 : they do not have to agree on an alternative @xmath88 that should replace @xmath87 ; they do not even have to know what alternatives @xmath137 the other members @xmath138 prefer .
see remark [ remark : core_story ] for further discussion . ]
[ remark : core ] the word `` core '' usually refers to the set of maximal alternatives with respect to some relation .
we adopt the word since the core without majority dissatisfaction is indeed the set of maximal elements with respect to the following extended dominance relation @xmath18 , which relates a subset @xmath71 of alternatives to an alternative @xmath87 .
can be seen as a dominance relation of a _ game in constitutional form _
@xcite , which associates a simple game with each pair of subsets of alternatives .
note , however , that the dominance relations ( e.g. , _ @xmath28-domination _ ) they actually analyze , unlike ours , require each player in a locally winning coalition to prefer _ all _ the alternatives in @xmath71 to @xmath87 . ] first , we extend @xmath28 s preference @xmath139 to a relation @xmath140 ( where @xmath141 is the power set of @xmath1 ) : @xmath142 if and only if there is @xmath143 such that @xmath144 .
to ones on its power set the set of opportunity sets . unless one is concerned with preferences for flexibility ( e.g. ,
* ) or freedom of choice , this approach is standard . ]
next , we extend the dominance relation @xmath145 to a relation @xmath146 : @xmath147 if and only if there is @xmath148 such that for all @xmath109 , @xmath142 . then , @xmath149 if and only if there is no @xmath150 such that @xmath147 ( if and only if @xmath151 ) . [ remark : core_story ] the core without majority dissatisfaction rejects any alternative ( `` status quo '' ) @xmath87 dominated by some set @xmath71 of alternatives with respect to the extended dominance relation in remark [ remark : core ] . according to this dominance relation , a coalition can block @xmath87 _ without having to agree on a replacement alternative_. admittedly , this notion of dominance may lack strong support , especially if one sticks to the usual interpretation of alternatives as complete descriptions of social states . however , when the standard solution ( the core ) selects too many alternatives , deleting some of them on a relatively weak ground could normatively be desirable .
the point of the main results is that they provide a condition ( nakamura s inequality ) ensuring that something remains even after alternatives are rejected on weak grounds .
having said that , we give an example where blocking behavior without agreeing on a single alternative @xmath88 is plausible .
one such example is a _
popularity contest among certain goods_. consider the problem of selecting the `` best '' articles published in economic journals in 2010 , for instance .
for simplicity , assume that @xmath152the opinions ( preferences ) of only three experts are elicited .
since one wants to compare different articles , a natural candidate for an alternative is an _ article _ , which is not a social state .
then the idea in remark [ remark : core ] of a set @xmath71 dominating an article @xmath87 should be intuitive enough , since it can be _ restated _ as follows : there are a majority , say the coalition @xmath153 , of experts @xmath28 and a feasible social state @xmath154 that dominates the feasible social state @xmath155 in the conventional sense : each @xmath156 prefers @xmath157 to @xmath155 ; that is , @xmath158 prefers @xmath159 to @xmath87 and @xmath160 prefers @xmath161 to @xmath87 . when one can assume that each allocation @xmath162 is feasible , this restatement makes a perfect sense . is a disposable private good _ that is available in sufficient quantity , the assumption is satisfied . note that we require a majority despite dealing with a private good , since we are considering a popularity contest .
_ if each @xmath143 is a public good _
, @xmath163 generally describes an _ infeasible _ , imaginary social state in which each @xmath28 consumes the public good @xmath132 .
the restatement loses some validity because of the lack of feasibility .
nevertheless , it does not lose all the validity in our view , since preferences are often elicited in a very incomplete manner in the real world , as footnote [ foot : core_scheme ] suggests . ] anyone accepting the conventional dominance relation defining the core would be ready to accept the extended dominance relation .
the following two lemmas are immediate from the definition .
the first one says that the core without majority dissatisfaction depends only on the set of maximal elements of each player .
[ c+maximal ]
let @xmath29 , @xmath164 be two profiles satisfying @xmath165 for all @xmath28
. then @xmath31 .
[ c+in_c ] for each profile @xmath29 , we have @xmath166 .
the inclusion @xmath167 is strict for some profile .
each of examples [ ex : core1 ] and [ ex : core2 ] below for graph representations of the profiles in these examples . ] shows that the inclusion is sometimes strict .
example [ ex : core1 ] also shows that _ an alternative can be in the core even if it is not maximal with respect to anyone s preference _ : [ ex : core1 ] let @xmath168 and let @xmath14 consist of the coalitions having a majority ( i.e. , having at least two players ) .
let @xmath169 .
define a profile by @xmath170 , @xmath171 , and @xmath172 .
then the sets @xmath173 of maximal elements of @xmath1 with respect to @xmath174 are given by @xmath175 , @xmath176 , and @xmath177 .
but the core @xmath178 is @xmath179 .
so , the core is neither a subset nor a superset of @xmath180 .
the core @xmath32 without majority dissatisfaction is @xmath181 , which is a subset of @xmath182 .
example [ ex : core2 ] also shows that _ the core , even if it is nonempty , does not necessarily intersect the union of the maximal elements of individual preferences _ : [ ex : core2 ] we modify the simplest voting paradox ( a cycle involving three alternatives and three players ) by adding an alternative @xmath131 .
let @xmath168 and let @xmath14 consist of the coalitions having a majority . let @xmath183 .
define a profile by @xmath184 , @xmath185 , and @xmath186 .
then the core @xmath178 is @xmath187 .
so it does not intersect @xmath188 .
the core @xmath32 without majority dissatisfaction is empty .
unlike the core , the core without majority dissatisfaction is always included in the union of the maximal elements of individual preferences : [ c+in_maximals ] for each profile @xmath29 , we have @xmath189 . shows that the inclusion is strict for some profile . ] by lemma [ c+in_c ] , it suffices to show that @xmath32 is a subset of @xmath190 .
suppose @xmath97 but @xmath191 .
then , @xmath192 for any @xmath193 .
this implies that @xmath194 for any winning coalition @xmath148 .
( such an @xmath195 exists since @xmath84 . ) by the definition of @xmath32 , we have @xmath196 .
we consider in the rest of the paper profiles _ for _ a set @xmath26 of alternatives , that is , measurable profiles consisting of _ preferences that have a maximal element of @xmath26_. we now give a version of nakamura s theorem for profiles _ for _ a set @xmath26 of alternatives : [ nakamura - max ] let @xmath105 be a simple game and @xmath20 an agenda .
let @xmath102 be the set of measurable profiles of preferences having a maximal element of @xmath26 .
then the following statements are equivalent:(ii ) is _ not _ the same as the following statement ( example [ ex : core2 ] ) : for each @xmath197 , if @xmath198 , then @xmath199 . ]
+ @xmath127 ; + the core @xmath200 without majority dissatisfaction is nonempty for all @xmath197 ; + the core @xmath110 is nonempty for all @xmath197 .
( i)@xmath201(ii ) .
suppose @xmath202 for some profile @xmath29 for @xmath26 . by definition [ c+ ] , for each @xmath97 , there is a winning coalition @xmath203 such that @xmath204 .
we claim that @xmath205 .
( otherwise , there is an @xmath28 who is in @xmath206 for all @xmath97 .
it follows that @xmath25 has no maximal element of @xmath26 . )
the claim shows that @xmath207 .
( ii)@xmath201(iii ) .
immediate from lemma [ c+in_c ] .
( iii)@xmath201(i ) .
suppose @xmath208 .
we construct a profile @xmath29 for @xmath26 such that @xmath209 .
let @xmath210 .
_ step _ 1 , _ case _
( a ) : @xmath211 is finite . _
we construct a profile @xmath29 such that the dominance relation @xmath18 has a cycle _ consisting of @xmath211 alternatives ( and the cycle contains an alternative @xmath212 greater than any alternative @xmath88 not belonging to the cycle ) . by the definition of the nakamura number
, there is a collection @xmath213 of winning coalitions such that @xmath214 .
choose a subset @xmath215 and write @xmath216 .
fix a cycle ( noting that @xmath217 means @xmath8 ) @xmath218 and a relation @xmath219 .
now , go to step 2 .
_ step _ 1 , _ case _
( b ) : @xmath211 is infinite .
_ we construct a profile @xmath29 such that the dominance relation @xmath18 defines an increasing transfinite sequence _ of length @xmath211 ( and the sequence contains an alternative @xmath212 greater than any alternative @xmath88 not belonging to the sequence ) . recall that if @xmath211 is an ordinal , then @xmath220 . by the definition of the nakamura number , there is a collectionsince @xmath221 is a ( well - orderable ) set whose cardinal number is @xmath211 , there is a bijection that maps each @xmath222 into an element @xmath223 of @xmath221 .
so we can write @xmath224 . ]
@xmath225 of winning coalitions @xmath223 such that @xmath226 .
choose a subset @xmath227 .
fix a relation@xmath228 which defines an increasing transfinite sequence of alternatives , does not end : if @xmath222 , then @xmath229 .
this is because @xmath211 , being a nakamura number , is a cardinal number and any infinite cardinal number is a limit ordinal by lemma [ lem : card - lim ] . ] and another relation @xmath219 .
_ step _ 2 .
we define a profile @xmath16 by specifying , for each pair @xmath230 , the set @xmath231 of players that prefer @xmath87 to @xmath88 .
note that @xmath232 .
( in the following , the letter @xmath33 denotes an ordinal number , including a natural number denoted @xmath233 in case ( a ) above . ) if @xmath234 , let @xmath235 . if @xmath236 , let @xmath237 .
if @xmath238 , let @xmath239 .
the profile @xmath29 is measurable since @xmath223 , @xmath15 , @xmath240 .
the profile @xmath29 is for @xmath26 ( i.e. , @xmath241 ) , since we can show that each @xmath28 s preference @xmath25 has a maximal element of @xmath26 .
for example , if @xmath242 , but @xmath243 for @xmath244 , then every alternative in @xmath245 except @xmath246 , @xmath247 , @xmath248 is a maximal element of @xmath25 .
( more formally , the set of maximal elements of @xmath25 is @xmath249 , which is nonempty since @xmath250 , where @xmath251 . )
the dominance relation @xmath18 is clearly @xmath252 , since @xmath223 is winning , @xmath15 contains a winning coalition , and @xmath42 is losing .
it follows that the core @xmath110 is empty .
it is instructive to compare the proof of the `` @xmath253 '' direction of theorem [ nakamura - thm ] with a direct proof of ( i)@xmath201(iii ) , which can be given in a way similar to that of ( i)@xmath201(ii ) above .
these proofs go as follows : suppose @xmath202 for some profile @xmath29 . for `` @xmath253 '' of theorem [ nakamura - thm ] ,
@xmath118 is a profile of _ acyclic _ preferences ; for ( i)@xmath201(iii ) , @xmath197 is a profile of preferences having a maximal element ( of @xmath26 ) .
then , + ( a ) for each @xmath97 , there are @xmath99 and a winning coalition @xmath203 such that @xmath254 ( hence @xmath255 ) .
+ note that @xmath87 is not a maximal element for the players in @xmath206 .
the desired conclusion @xmath207 follows if we show @xmath205 .
so suppose there is an @xmath28 who is in @xmath206 for all @xmath97 .
* to show `` @xmath253 '' of theorem [ nakamura - thm ] , assume @xmath26 is finite . by repeated application of ( a )
, we find the dominance @xmath256 contains an infinitely ascending sequence : @xmath257 .
since @xmath26 is finite , the sequence contains a cycle such as @xmath258 .
it follows that @xmath28 s preference contains the same cycle , which violates the assumption . * for ( i)@xmath201(iii ) of theorem [ nakamura - max ] , a contradiction is immediate : let @xmath97 be a maximal element for @xmath25 ; then @xmath259 is violated .
we do not even need the assumption that @xmath26 is finite .
we remark that the possibility remains that @xmath259 if @xmath97 is not a maximal element for @xmath28 .
so @xmath28 s preference may contain a cycle without contradiction ; it consists only of non - maximal elements for her .
the following example is an application of theorem [ nakamura - max ] .
it gives a condition for an _ infinite _ set of alternatives to have an element in the core , that is , a maximal element with respect to the dominance relation @xmath18 .
let @xmath260 $ ] be the unit interval on the set of real numbers and let @xmath37 consist of all subsets of @xmath15 .
define a simple game @xmath14 by @xmath107 if and only if @xmath261 is countable .
then , it is easy to show that @xmath35 is uncountable . let @xmath1 be a countable set of alternatives ( e.g. , the set of rational numbers in @xmath262 $ ] ) .
suppose that for each @xmath28 , her preference @xmath25 has a maximal alternative ( e.g. , a utility function representing @xmath25 has finitely many `` peaks , '' all corresponding to rational numbers ) .
then , theorem [ nakamura - max ] implies that @xmath263 and @xmath19 both contain alternatives . the profiles constructed in the proof of ( iii)@xmath201(i ) of theorem [ nakamura - max ] consist of individual preferences that may have more than one maximal alternative .
however , we can modify the proof in such a way that each individual preference has _ exactly one _ maximal alternative . that gives the following proposition : [ nakamura - max2 ] let @xmath105 be a simple game and @xmath20 an agenda .
then the three equivalent statements , , and in theorem [ nakamura - max ] are equivalent to the following statements : + the core @xmath200 without majority dissatisfaction is nonempty for all measurable profiles @xmath29 of preferences having _ exactly one _ maximal element of @xmath26 .
+ the core @xmath200 without majority dissatisfaction is nonempty for all measurable profiles @xmath29 of linear orders having a maximal element of @xmath26 .
+ the core @xmath110 is nonempty for all measurable profiles @xmath29 of preferences having _ exactly one _ maximal element of @xmath26 . + the core @xmath110 is nonempty for all measurable profiles @xmath29 of linear orders having a maximal element of @xmath26 . ( ii)@xmath201(ii.a)@xmath201(ii.b ) and ( iii)@xmath201(iii.a)@xmath201(iii.b )
are obvious .
( ii.b)@xmath201(iii.b ) is immediate from lemma [ c+in_c ] .
( iii.b)@xmath201(i ) .
suppose @xmath208 .
we construct a profile @xmath29 satisfying the condition such that @xmath209 .
let @xmath210 . in _ step _ 1 of the proof of theorem [ nakamura - max ] ,
replace the relation @xmath264 by an arbitrary asymmetric subrelation @xmath265 that is linear on @xmath266 and satisfies @xmath267 whenever @xmath268 and @xmath269 .
we replace _ step _ 2 of the proof by the following argument .
_ case _ : @xmath211 is finite . define @xmath270 and for all @xmath271 , @xmath272 then @xmath273 is a family of ( possibly empty ) pairwise disjoint coalitions in @xmath37 such that @xmath274 for all @xmath233 and @xmath275 ( @xmath193 is in the first @xmath276 such that @xmath243 ) .
define @xmath29 as follows : for each @xmath233 , all players @xmath28 in @xmath276 have the same linearly ordered preference @xmath25 with maximal element @xmath277 , obtained by taking the transitive closure of @xmath278 .
obviously , @xmath29 is measurable since for each @xmath87 , @xmath95 , @xmath231 is a finite union of @xmath276 s .
also , @xmath18 includes @xmath279 .
( if @xmath280 , we have @xmath281 . if @xmath236 , then @xmath237 . )
it follows that @xmath282 .
_ case _ : @xmath211 is infinite .
note that @xmath283 .
define @xmath29 as follows : if @xmath236 , then @xmath237 . if @xmath284 , then @xmath239 . if @xmath285 , @xmath286 if and only if either , @xmath287 is @xmath28 s _ preferred class _ of alternatives in @xmath245 and @xmath288 is her _ less - preferred class_. her preference @xmath25 is linear on @xmath245 satisfying the following conditions : ( a ) if two alternatives belong to her preferred class , then @xmath28 prefers the one with the smaller index ; ( b ) @xmath28 prefers each alternative in her preferred class to any in her less - preferred class ; ( c ) if two alternatives belong to her less - preferred class , then @xmath28 prefers the one with the greater index .
for example , if @xmath242 , but @xmath289 for @xmath290 , then @xmath28 ranks the alternatives in the following order : @xmath212 , @xmath291 , @xmath292 , @xmath293 , @xmath294 , , @xmath248 , @xmath247 , @xmath246 . ] * @xmath295 and @xmath64 ; or * @xmath296 and @xmath297 ; or * @xmath298 and @xmath299 .
this is equivalent to saying that iff either ( 1 ) @xmath64 and @xmath300 or ( 2 ) @xmath299 and @xmath301 .
note that the first two cases can be restated : if @xmath64 , then @xmath302 and @xmath303 . ]
@xmath304 the profile @xmath29 is measurable since @xmath15 , @xmath42 , @xmath305 , @xmath306 .
clearly , each @xmath28 s preference @xmath25 is a linear order on @xmath1 that has a unique maximal element of @xmath26 , namely the alternative in her preferred class @xmath287 with the smallest index .
the dominance relation @xmath18 clearly includes @xmath279 , since @xmath307 for each @xmath308 , for example .
it follows that @xmath282 .
the argument for the finite @xmath211 case of the proof does not go through for the infinite case , since @xmath276 does not necessarily belong to @xmath37 when @xmath233 is infinite .
the argument for the infinite case causes difficulty when applied to the finite case . for example , suppose @xmath309 and @xmath310 .
since @xmath311 and @xmath312 , we have @xmath313 by the definition of @xmath29 . since @xmath314 and @xmath315 , we have @xmath316 . since @xmath317
, @xmath25 is not asymmetric .
we extend the framework by introducing a collection @xmath36 , consisting of subsets of the set @xmath15 of players . recall that @xmath37 consists of the _
coalitions_intuitively , they are the observable or identifiable sets of players . in contrast , @xmath36 consists of the sets _ for which one can assign winning / losing status_sometimes they are the sets whose size is well - defined ( example [ borel - lebesgue ] ) ; other times they are the sets that are half - identifiable or listable ( example [ recursive - re ] ) .
we assume @xmath318 , which means that one can assign such a status for any coalition .
a * collection @xmath221 of winning sets * is a subcollection of @xmath36 satisfying @xmath319 and @xmath320 .
given @xmath221 , the most natural simple game one can define is the following : the * @xmath37-simple game @xmath14 induced by @xmath221 * consists of the winning coalitions ( winning sets that are also coalitions ) , that is , @xmath321 .
we can define the _ core _ and the _ core without majority dissatisfaction _ as before , with @xmath14 replaced by @xmath221 .
lemma [ c+in_c ] and @xmath322 imply the following .
[ bet - c+c ] let @xmath323 and @xmath324 .
then the following statements are true : + @xmath325 .
+ @xmath326 .
since we had better be able to identify who prefers a given alternative to another , we leave the definitions of _ measurable _ profiles and _ profiles for _ an agenda unaltered .
we have assumed @xmath327 in the previous sections , but there are reasons for distinguishing the collection @xmath36 from the boolean algebra @xmath37 of coalitions .
we illustrate that by two examples in this section .
we also give an example of a decision - making situation where the framework makes sense .
the following example gives some , though limited , justification for the extended framework : [ borel - lebesgue ] let @xmath260 $ ] be the unit interval on the set of real numbers , @xmath37 the @xmath328-algebra of borel sets ( i.e. , @xmath37 is the smallest @xmath328-algebra containing all open sets in @xmath262 $ ] ) , @xmath36 the @xmath328-algebra of lebesgue measurable sets , and @xmath329 lebesgue measure .
let @xmath323 be any collection of winning sets defined in terms of the measure alone .
( for example , @xmath221 consists of the sets @xmath330 satisfying @xmath331 .
alternatively , @xmath221 consists of the sets @xmath330 satisfying @xmath332 . )
we do have @xmath333 , though this fact is not often exploited ..
@xcite , when restricted to our framework , focus on simple games @xmath105 , rather than on @xmath39 . ] observe that for each @xmath334 there is @xmath335 such that @xmath336 and @xmath337 .. let @xmath338 .
apply the following proposition @xcite and let @xmath339 : if @xmath340 $ ] is any set , then there is a borel set @xmath341 such that @xmath342 and @xmath343 , where @xmath344 is lebesgue outer measure . ]
the last observation in example [ borel - lebesgue ] explains why it fails to give a compelling justification for the extended framework , because of the following lemma .
the lemma suggests that under a certain condition , it is not very meaningful to introduce a collection @xmath39 , even when it makes sense to extend @xmath37 to @xmath36 .
the proof is straightforward .
let @xmath323 and @xmath324 .
suppose that for each winning set @xmath345 , there exists a winning coalition @xmath107 satisfying @xmath336 . then @xmath346 and @xmath347 . now we give a more compelling justification for the framework in which @xmath36 and @xmath221 are introduced : [ recursive - re ] let @xmath348 be the set of natural numbers .
a natural and faithful way to model _
identifiable _ or _ half - identifiable _ sets of players is to let @xmath37 be the algebra of recursive sets ( coalitions ) and @xmath36 the lattice of r.e .
is recursive if and only if @xmath67 and @xmath349 are both r.e .
@xcite gives detailed discussion of recursion theory ( computability theory ) .
the papers by @xcite contain short reviews of recursion theory . ]
the _ first reason _ for introducing @xmath350 in our framework is that we can not let @xmath37 be the lattice of r.e .
sets , since the lattice is not a boolean algebra .
the _ second reason _ is that the natural system @xmath351 for indexing r.e .
sets is easier to handle than the system @xmath352 that can be used for indexing recursive sets ( where @xmath353 is the @xmath354th partial recursive function and @xmath355 its domain ) .
for example , while @xmath353 corresponds to ( i.e. , is the characteristic function of ) no recursive set for some @xmath356 ( and there is no algorithm to decide whether a given @xmath354 corresponds to some recursive set ) , @xmath355 corresponds to ( i.e. , is ) an r.e .
set for any @xmath356 .
one can therefore write any class of r.e .
sets as @xmath357 for some ( not necessarily r.e . ) set @xmath358 of indices , without worrying that some @xmath359 might correspond to no r.e .
@xcite gives more reasons for preferring @xmath351 to @xmath352 .
we now give a _ reason for introducing @xmath323 _ into our framework .
we exhibit @xmath221 , @xmath321 , and @xmath29 for which @xmath360 .
a set is _ cofinite _ if it is the complement of a finite set ; otherwise , it is _ coinfinite_. a coinfinite r.e
. set @xmath361 is * * maximal coinfinite * * if it has only trivial supersets : if @xmath195 is a coinfinite r.e .
set satisfying @xmath362 , then @xmath363 is finite .
let @xmath364 be the collection of all maximal coinfinite sets . since maximal coinfinite sets are nonrecursive , we have @xmath365 , which is not a simple game according to our definition
. we can nevertheless conclude @xmath366 and define the core , obtaining @xmath367 for any profile @xmath29 .
let @xmath368 and define a profile @xmath29 by @xmath369 for all @xmath193 .
then , @xmath370 .
let @xmath371 .
then we can easily show that @xmath372 consists of all cofinite sets ; therefore , @xmath373 . let @xmath374 be a countable set .
let @xmath375 be a maximal coinfinite set .
define a profile @xmath29 as follows : @xmath376 if @xmath377 ; @xmath378 otherwise .
then , @xmath231 is @xmath379 if @xmath380 for @xmath377 and @xmath381 ; it is @xmath42 otherwise .
so @xmath29 is measurable and each player has a maximal alternative .
we have @xmath382 . also , @xmath383 is @xmath361 if @xmath384 ; it is @xmath42 otherwise .
so , @xmath385 but @xmath386 .
we finally give an example of a decision - making setting to which the framework and example [ recursive - re ] can be applied .
our example is one of medical treatment , but it also represents other examples of _ multi - criterion decision making _ such as court rulings and facility location problems . consider a decision support system that helps medical professionals by selecting a reasonable set of alternatives from which they can choose .
such a system can be regarded as a _ solution _ ( like the core ) which assigns a set of alternatives to each profile of preferences . here , an _ alternative _ describes , for instance , what and how much medicine to prescribe to the patient and when and what operation to perform on her .
we assume that there are countably many alternatives .
this is a natural assumption if the system is implemented on a digital computer .
a _ player _ in the framework is understood to be a _ criterion _ such as sex , age , temperature , blood pressure , or the result of a medical examination .
a criterion may be a combination of some of these .
we assume that there are countably many criteria .
the infinite criterion model is appropriate if one can not tell in advance how many criteria one needs to evaluate to reach a decision . to formalize the idea that _ coalitions _ are identifiable
, we assume that each coalition is _ recursive _ ; that is , there is an algorithm to decide whether each criterion belongs to the coalition .
when @xmath28 is a criterion , its _ preference _ is a provisional judgment interpreted as follows : @xmath93 if and only if `` alternative @xmath87 is better than @xmath88 in terms of criterion @xmath28 for the patient in question . ''
so , whether @xmath108 is true is determined only after a patient is given .
we assume that a profile is recursive : there is an algorithm for deciding ( for all @xmath87 , @xmath88 , and @xmath28 ) whether @xmath108 .
( it follows that profiles are measurable . )
this formalizes the idea that the system should give an answer at least for those profiles that it can grasp .
we want the decision support system to eliminate the alternatives that are , for `` almost all criteria , '' worse than some other alternatives in an agenda @xmath26 . in other words
, we want to eliminate alternatives @xmath87 such that the set @xmath387 contains `` almost all criteria . '' as a suitable solution for that purpose , we can adopt the core without majority dissatisfaction .
we also formalize the intuition that a set @xmath388 * contains `` almost all criteria '' * as follows : @xmath389 for some _ maximal coinfinite _ set @xmath361 ( defined in example [ recursive - re ] ) . according to this formalization ,
if @xmath390 contains `` almost all criteria , '' then ( it does not necessarily mean that it contains all but finitely many criteria , but ) one can not algorithmically generate infinitely many criteria not belonging to this set ( appendix [ maximal_compl_no_re ] ) .
suppose , for the time being , that one can assign winning / losing status only to ( recursive ) coalitions .
one can naturally define that a coalition @xmath195 is * winning * if it contains `` almost all criteria '' ; that is , @xmath362 for some maximal coinfinite @xmath361 . _
a problem is that the set @xmath387 may contain `` almost all criteria '' without containing a winning coalition .
that is , we want to eliminate @xmath87 , but the definition ( definition [ c+ ] ) of the core without majority dissatisfaction does not allow us to eliminate it . _ indeed , example [ recursive - re ] shows that when the set itself is maximal coinfinite , it contains no winning coalition . to rectify this problem , we assign winning / losing status to every r.e .
set of criteria .
( an _ r.e .
set _ is half - identifiable in the sense that there is an algorithm that enumerates its elements . )
we define that an r.e .
set is * winning * if it contains `` almost all criteria '' as above .
we then have the equivalence of the following for r.e .
sets @xmath390 : ( i ) @xmath390 contains `` almost all criteria '' ; ( ii ) @xmath390 contains a winning r.e . set ; ( iii ) @xmath390 is winning . fortunately , when @xmath26 is r.e . , the set @xmath387 is also r.e .
, since it can be rewritten as @xmath391 \}$ ] .
so to determine whether to reject @xmath87 here , we only need to check whether this set is winning .
before stating the main result for the extended framework , we need to extend the notion of the nakamura number .
let @xmath350 be a collection that includes @xmath392 , the boolean algebra of coalitions .
let @xmath39 be a collection of winning sets .
a nonempty collection @xmath393 is a _
( @xmath37)-cover _ of @xmath334 if @xmath394 .
if @xmath395 is a finite cover , then @xmath396 , since @xmath37 is a boolean algebra .
so , we can assume without loss of generality that @xmath397 is either @xmath158 or infinite .
let @xmath398 be the collection of pairs @xmath399 such that + ( a ) @xmath400 and + ( b ) @xmath401 is a correspondence that maps each winning set @xmath402 to a @xmath37-cover @xmath403 of @xmath404 and that satisfies @xmath405 .
[ kappa ] the * kappa number * @xmath38 of a collection @xmath221 of winning sets relative to @xmath37 is @xmath406 ( greater than any cardinal number ) if @xmath407 ; otherwise , it is the cardinal number given by , where each @xmath408 is a cardinal number .
it suffices to show that @xmath33 is an initial ordinal .
suppose @xmath409 .
then , by the definition of the supremum , @xmath410 for some @xmath411 . since @xmath408 is a cardinal , @xmath412 .
but @xmath413 implies that @xmath33 is not equipotent to @xmath57 . ]
@xmath414 there is another , obvious extension @xmath415 ( defined for collections of winning sets ) of the nakamura number @xmath211 ( defined for simple games ) .
let @xmath416 be defined exactly as before , with @xmath14 replaced by @xmath221 .
[ nu - kappa - nu ] @xmath417 for @xmath324 .
@xmath418 is a consequence of the assumptions @xmath319 and @xmath320 .
we show @xmath419 next .
if @xmath420 , the result is obvious .
so , suppose otherwise . by definition [ kappa ] , there exists @xmath421 satisfying ( a ) and ( b ) in the definition of @xmath398 such that @xmath422 since @xmath423 for all @xmath424 , we have @xmath425 .
it follows from the definition of @xmath416 that @xmath426 .
we show @xmath427 finally .
suppose @xmath428 as above .
then , by the definition of @xmath211 , there is a collection @xmath429 such that @xmath430 and @xmath431 . for each @xmath402 ,
let @xmath432 .
we claim that @xmath421 .
( ( a ) @xmath433 .
( b ) @xmath403 is a cover of @xmath404 since @xmath434 . also , @xmath435 . )
it follows from definition [ kappa ] that @xmath436 .
the inequalities in lemma [ nu - kappa - nu ] are sometimes strict . in the following example
, we have @xmath437 .
[ nu - kappa - nu2 ] consider @xmath260 $ ] , the unit interval on the set of real numbers .
let @xmath36 be the collection of all subsets of @xmath15 .
let @xmath221 be the collection of all dense subsets @xmath438 of @xmath15 , that is , between any two distinct points in @xmath15 , there is an element of @xmath438 .
for example , both @xmath439 and @xmath440 belong to @xmath221 .
we have @xmath441 , since @xmath442 .
let @xmath37 be the collection of all finite or cofinite sets @xmath443 .
let @xmath14 be the collection of all cofinite sets @xmath443 .
we prove that @xmath444 .
( @xmath167 ) : suppose @xmath107
. then , @xmath445 .
since @xmath261 is finite , @xmath195 is a dense subset of @xmath15 .
( @xmath446 ) : suppose @xmath447 .
since @xmath335 , either @xmath195 is finite or cofinite . if @xmath195 is finite , it is not dense in @xmath15 .
so @xmath195 must be cofinite .
we prove that @xmath448 , the cardinality of the continuum . by lemma [ nakamura - bounds ] , it suffices to show that @xmath449 for a collection @xmath450 implies @xmath451 . taking the complement , @xmath452 ( * ? ? ?
* theorem 1.3 , page 188 ) .
since the supremum is at most @xmath60 ( because @xmath453 is finite ) , we have @xmath451 .
finally , we prove that @xmath454 .
we first show that @xmath455 .
let @xmath456 , where @xmath404 , @xmath457 are the winning sets given above .
let @xmath458 and @xmath459 .
it is straightforward to show that @xmath460 . since @xmath461 and @xmath462 , we have @xmath455 . we next show that @xmath38 is not finite .
suppose it is finite .
then , there exists @xmath421 such that @xmath463 is finite and @xmath464 is finite for each @xmath465 .
since @xmath37 is a boolean algebra , this implies that @xmath466 , which in turn implies that @xmath467 is either finite or cofinite .
suppose @xmath467 is finite for some @xmath468 .
then @xmath469 implies that @xmath470 is finite , a contradiction .
hence @xmath467 is cofinite for all @xmath465 .
being a finite intersection of cofinite sets , @xmath471 is nonempty , contrary to assumption .
[ kappa - extends - nu ] @xmath472 is an extension of @xmath211 .
that is , if @xmath473 , then @xmath474 .
suppose @xmath475 .
then , @xmath476 .
lemma [ nu - kappa - nu ] implies @xmath477 .
since @xmath415 extends @xmath211 and @xmath221 is a simple game in this case , we have @xmath478 , from which the conclusion follows .
we now give the main result for the extended framework .
[ nakamura - max - extended ] let @xmath39 be a collection of winning sets and @xmath20 an agenda .
let @xmath102 be the set of measurable profiles of preferences having a maximal element of @xmath26 .
then the following two statements are equivalent : + @xmath479 ; + the core @xmath480 without majority dissatisfaction is nonempty for all @xmath197 .
+ moreover , these equivalent statements imply , but are not implied by + the core @xmath481 is nonempty for all @xmath197 . and [ nu - kappa - nu ]
are not useful for proving theorem [ nakamura - max - extended ] from theorem [ nakamura - max ] , and vice versa . ]
( i)@xmath201(ii ) .
suppose @xmath482 for some profile @xmath197 .
we show that @xmath483 . by the definition ( definition [ c+ ] ) of @xmath32 , for each @xmath97
, there is a winning set @xmath484 such that @xmath485 .
we claim that @xmath486 .
( otherwise , there is a player @xmath28 whose preference has no maximal element of @xmath26 . )
let @xmath487 .
we have @xmath488 . for each @xmath489 , let @xmath490 we have @xmath491 .
we claim that @xmath421 .
( _ details_. we verify ( b ) of the definition of @xmath398 .
first , @xmath492 since @xmath29 measurable implies @xmath493 .
second , @xmath494 is a cover of @xmath495 since @xmath496 .
third , @xmath497 by the claim above . ) by the definition ( definition [ kappa ] ) of @xmath472 , we get @xmath498 ( ii)@xmath201(i ) .
suppose @xmath499 .
we construct a profile @xmath241 such that @xmath482 . by the definition ( definition [ kappa ] ) of @xmath500
, there exists @xmath421 such that and @xmath501 in the form below ( footnote [ note : indexed - collection ] ) . ] + ( a ) @xmath502 , where @xmath503 : + ( b ) @xmath395 maps each @xmath504 to @xmath505 ( where @xmath506 ) satisfying @xmath507 and @xmath508 ; + ( c ) @xmath509 .
write @xmath510 and let @xmath511 .
_ case _ : @xmath472 is finite .
we construct a profile @xmath29 such that the dominance relation @xmath18 has a cycle consisting of @xmath472 alternatives . since @xmath512 is finite for all @xmath513 in this case
, @xmath514 is a finite union of elements of the boolean algebra @xmath37 .
so we can assume @xmath515 and @xmath516 without loss of generality .
by ( c ) , we have @xmath517 .
write @xmath518 and fix a relation @xmath219 .
let @xmath519 for all @xmath193 .
the profile @xmath29 is the same as that in the proof of theorem [ nakamura - max ] , except that @xmath223 is replaced by @xmath520 .
the rest of the proof runs as before .
_ case _ : @xmath472 is infinite .
we construct a profile @xmath29 such that for each alternative @xmath521 , there is a winning set of players @xmath28 who prefer another alternative @xmath522 to @xmath523 . with smaller index to @xmath523 ( if @xmath524 , then @xmath525 ) . ]
since preferences involving alternatives outside @xmath26 are irrelevant , we construct the profile in such a way that it satisfies @xmath526 for all @xmath28 .
we define @xmath29 by specifying @xmath231 for each pair @xmath527 satisfying @xmath528 .
( if @xmath529 does not satisfy the conditions concerning the indices , then @xmath239 . ) note that each such pair can be uniquely written as @xmath530 for some @xmath531 .
is the unique ordinal isomorphic to @xmath532 .
] let @xmath533 if @xmath516 ; otherwise , we can assume @xmath534 is infinite . define @xmath29 by the following , which immediately establishes that @xmath29 is measurable : for a graph representation of the profile .
for the subsequent argument , we need to verify that @xmath535 and @xmath523 corresponding to the first two cases do exist in @xmath26 , that is , @xmath536 ( and @xmath537 , which is obvious ) . for the first and the third cases , since @xmath537
, we have @xmath538 .
this is because @xmath539 , being an infinite cardinal number , is a limit ordinal by lemma [ lem : card - lim ] . for the second case , since @xmath540
, we have @xmath541 .
this is because @xmath539 , being a cardinal number , is not equipotent to @xmath537 .
similarly , since @xmath542 , we have @xmath543 . if @xmath33 and @xmath57 are finite , the conclusion is obvious since @xmath539 is infinite in this case .
suppose otherwise .
then , by lemma [ lem : sumsum ] , we have @xmath544 .
this implies that there is no one - to - one function from @xmath539 to @xmath545 .
therefore , @xmath536 . ]
@xmath546 for each @xmath521 , we have @xmath547 , which equals @xmath548 ( if @xmath513 ; the equality holds whether @xmath534 is @xmath158 or infinite ) or @xmath15 ( otherwise ) . in either case , the set contains a winning set @xmath549 . therefore , @xmath550 .
this establishes that @xmath482 . to establish that @xmath241
, we need to show that each @xmath28 s preference @xmath25 has a maximal element of @xmath26 .
since @xmath551 , we have @xmath552 .
so , for each @xmath193 , there is an @xmath513 such that @xmath553 . since @xmath554 for any @xmath555
, we have @xmath556 by the definition of @xmath29 .
( ii)@xmath201(iii ) .
immediate from lemma [ bet - c+c ] .
( iii)@xmath557(i ) .
consider example [ nu - kappa - nu2 ] .
we first prove that @xmath558 for all @xmath241 and all @xmath20 . by lemma [ bet - c+c ] , it suffices to show that @xmath559 .
suppose @xmath560 is not in @xmath41 . then , there are @xmath561 and @xmath562 such that @xmath563 . since @xmath29 is measurable , @xmath564 is either finite or cofinite . if it is finite , then @xmath565 is finite , a contradiction .
it follows that @xmath566 is cofinite , hence it belongs to @xmath14 .
so @xmath567 .
choose an agenda @xmath26 satisfying @xmath568 .
( i ) is violated since @xmath569 . on the other hand , since @xmath570 , theorem [ nakamura - max ] implies that @xmath571 for all @xmath197 .
then ( iii ) is satisfied since @xmath558 .
29 natexlab#1#1url # 1`#1`urlprefix ambrus , a. , rozen , k. , jul .
2008 . revealed conflicting preferences .
discussion paper 1670 , cowles foundation , yale university , new haven .
andjiga , n. g. , mbih , b. , 2000 . a note on the core of voting games .
journal of mathematical economics 33 , 367372 .
andjiga , n. g. , moulen , j. , 1989 .
necessary and sufficient conditions for @xmath572-stability of games in constitutional form . international journal of game theory 18 , 91110 .
andjiga , n. g. , moyouwou , i. , 2006 . a note on the non - emptiness of the stability set when individual preferences are weak orders . mathematical social sciences 52 , 6776 .
arrow , k. j. , 1963 . social choice and individual values , 2nd edition .
yale university press , new haven .
austen - smith , d. , banks , j. s. , 1999 .
positive political theory i : collective preference .
university of michigan press , ann arbor .
banks , j. s. , 1985 .
sophisticated voting outcomes and agenda control . social choice and welfare 1 , 295306 .
banks , j. s. , 1995 .
acyclic social choice from finite sets . social choice and welfare 12 , 293310 .
banks , j. s. , duggan , j. , le breton , m. , 2006 . social choice and electoral competition in the general spatial model .
journal of economic theory 126 , 194234 .
dasgupta , p. , maskin , e. , 2008 . on the robustness of majority rule .
journal of the european economic association 6 , 949973 .
duggan , j. , 2007 . a systematic approach to the construction of non - empty choice sets .
social choice and welfare 28 , 491506 .
fishburn , p. c. , 1970 .
arrow s impossibility theorem : concise proof and infinite voters .
journal of economic theory 2 , 1036 .
friedberg , r. m. , 1958 .
three theorems on recursive enumeration . i. decomposition , ii .
maximal set , iii .
enumeration without duplication .
journal of symbolic logic 23 , 309316 .
hrbacek , k. , jech , t. , 1984 .
introduction to set theory , second , revised and expanded edition .
marcel dekker , new york .
kalai , g. , rubinstein , a. , spiegler , r. , 2002 . rationalizing choice functions by multiple rationales .
econometrica 70 , 24812488 .
kreps , d. m. , 1979 . a representation theorem for `` preference for flexibility '' .
econometrica 47 , 565577 .
kumabe , m. , mihara , h. r. , 2008 .
computability of simple games : a characterization and application to the core .
journal of mathematical economics 44 , 348366 .
kumabe , m. , mihara , h. r. , 2008 .
the nakamura numbers for computable simple games . social choice and welfare 31 , 621640 . le breton , m. , salles , m. , 1990 .
the stability set of voting games : classification and genericity results .
international journal of game theory 19 , 111127 .
lipman , b. l. , 1991 .
how to decide how to decide how to : modeling limited rationality .
econometrica 59 , 11051125 .
martin , m. , merlin , v. , 2006 . on the characteristic numbers of voting games .
international game theory review 8 , 643654 .
mihara , h. r. , 1997 .
arrow s theorem and turing computability . economic theory 10 , 25776 .
mihara , h. r. , 1999 .
arrow s theorem , countably many agents , and more visible invisible dictators
. journal of mathematical economics 32 , 267287 .
mihara , h. r. , 2000 .
coalitionally strategyproof functions depend only on the most - preferred alternatives . social choice and welfare 17 , 393402 .
nakamura , k. , 1979 .
the vetoers in a simple game with ordinal preferences
. international journal of game theory 8 , 5561 .
odifreddi , p. , 1992 .
classical recursion theory : the theory of functions and sets of natural numbers .
elsevier , amsterdam .
penn , e. m. , 2006 .
the banks set in infinite spaces . social choice and welfare 27 , 531543 .
royden , h. l. , 1988 . real analysis ,
3rd edition .
macmillan , new york .
rubinstein , a. , 1980 .
stability of decision systems under majority rule .
journal of economic theory 23 , 150159 .
truchon , m. , 1995 .
voting games and acyclic collective choice rules .
mathematical social sciences 29 , 165179 .
this is supplementary material for `` preference aggregation theory without acyclicity : the core without majority dissatisfaction '' by masahiro kumabe and h. reiju mihara .
suppose @xmath33 is an infinite cardinal number that is not a limit ordinal .
then , being a successor ordinal , @xmath573 for some ordinal @xmath57 .
clearly , @xmath409 .
since @xmath33 is initial , there is no bijection between @xmath574 and @xmath57 .
but we can construct such a bijection @xmath575 as follows : @xmath576 if @xmath577 ; @xmath578 if @xmath579 ; @xmath580 if @xmath581 but @xmath582 . pick disjoint well - ordered sets @xmath583 and @xmath584 isomorphic to ordinals @xmath33 and @xmath57 , respectively . then
, we have @xmath585 ( * ? ? ?
* theorem 5.5 , page 152 ) .
this is equal to @xmath586 by the definition of the cardinal sum .
we `` replicate '' sufficiently many times an example for which @xmath32 is a proper subset of @xmath178 , and then add an `` insignificant '' player whose maximal alternative contains an alternative belonging to the difference @xmath588 .
we build on example [ ex : core1 ] here .
let @xmath589 and let @xmath14 consist of the coalitions having a majority .
let @xmath169 .
define a profile by @xmath590 , @xmath591 , @xmath592 , and @xmath593 ( or any preference such that @xmath131 is a maximal ) . then , as before , the core @xmath178 is @xmath179 and the core @xmath32 without majority dissatisfaction is @xmath181 .
so @xmath32 is a proper subset of @xmath594 .
let @xmath596 be an infinite r.e .
then @xmath597 is an r.e .
set containing @xmath361 . by the definition of maximal coinfinite sets ,
either @xmath363 is finite or @xmath195 is cofinite .
but the first case can not occur since @xmath598 is infinite . since @xmath195 is cofinite
, we have @xmath599 for some finite @xmath600 .
it follows that @xmath601 .
so @xmath595 is r.e . , implying that @xmath361 is recursive .
this contradicts the fact that @xmath361 is maximal coinfinite .
[ [ profile - for - the - proof - iirightarrowi - of - theoremnakamura - max - extended - when - kappa - is - infinite ] ] profile for the proof ( ii)@xmath201(i ) of theorem [ nakamura - max - extended ] , when @xmath472 is infinite ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
acyclicity of individual preferences is a minimal assumption in social choice theory .
we replace that assumption by the direct assumption that preferences have maximal elements on a fixed agenda .
we show that the core of a simple game is nonempty for all profiles of such preferences if and only if the number of alternatives in the agenda is less than the nakamura number of the game .
the same is true if we replace the core by the _ core without majority dissatisfaction _
, obtained by deleting from the agenda all the alternatives that are non - maximal for all players in a winning coalition . unlike the core , the core without majority dissatisfaction depends only on the players sets of maximal elements and is included in the union of such sets .
a result for an extended framework gives another sense in which the core without majority dissatisfaction behaves better than the core .
_ journal of economic literature _
classifications : c71 , d71 , c02 .
_ keywords : _ core , nakamura number , kappa number , simple games , voting games , maximal elements , acyclic preferences , limit ordinals .
|
the existence of dark matter was discovered in galaxy clusters where the velocity dispersions are nearly an order of magnitude higher than expected from the gravitational binding provided by the stellar masses of their visible galaxies ( @xcite ) .
clusters are gravitationally bound , quasi - equilibrium systems assembled over a hubble time via the infall of the mass in the surrounding field ( @xcite ) , which implies that both the dark matter and the galaxies within clusters have their origins in the field . because clusters are large systems that draw their mass and galaxy content from regions 20 across , measurements of cluster @xmath1 values should be representative of the field value , although not necessarily identical to it because of differential galaxy evolution .
the product of the field @xmath1 with the field luminosity density , @xmath17 , is equal to the mean mass density of the universe , @xmath18 ( @xcite ) .
the cosmological density parameter , @xmath19 , is therefore estimated as the ratio of the cluster @xmath1 ( corrected to the field ) to the @xmath20 for closure ( @xcite ) .
the resulting @xmath21 estimate has no dependence on @xmath22 for dynamically measured cluster masses ( @xcite ) .
the purpose of this paper is mainly to correct our cluster virial mass - to - light ratio , @xmath0 , ( carlberg , 1996 , hereafter @xcite ) to the field @xmath1 .
cluster dynamical masses are usually calculated from the virial mass estimator , @xmath11 , which has the drawback that it makes explicit assumptions about the dynamical state of the clusters .
its reliability is critically dependent on the galaxy population being in dynamical equilibrium with the cluster potential and the galaxy distribution having the same spatial distribution as the total mass distribution of the cluster .
measured virial mass - to - light ratios of clusters , @xmath0 , are generally in the range @xmath23 ( @xcite ) which , in ratio to @xmath24 ( @xcite ) indicates @xmath25 .
cluster @xmath26 values have not generally been accepted as conclusive because there are a number of uncontrolled sources of possible systematic error .
for instance , cluster galaxies with blue colors are known to have a higher velocity dispersion and are more extended than the red galaxies ( @xcite ) leading to virial mass estimates that are about a factor of four different for our dataset .
the substantial color difference between cluster and field galaxies opens up the possibility that their luminosities per unit mass are quite different . the assumptions which go into the cluster @xmath21 calculation need to be tested for their validity for a specific sample .
the possibility that the mass distributions of galaxy clusters are more extended than their constituent galaxy populations has been recognized for many years ( @xcite ) .
the detection problem is that dynamical mass estimators do not have any useful sensitivity to cluster mass outside the orbits of the galaxies .
a specific situation which leads to this bias is found in n - body simulations of cluster buildup in an @xmath27 cosmology with a realistic amount of substructure ( @xcite ) . in this case
the simulated `` galaxies '' become more concentrated within the cluster than its mass .
that is , the virial mass calculated from the `` galaxies '' systematically underestimates the total mass of the cluster in these simulations by as much as order of magnitude .
there are some observational indications of the possibility of a higher @xmath21 than that calculated from virial masses .
detailed mass modeling for the coma cluster gives a projected mass of @xmath28 within @xmath29 of the center ( with a normal mass - to - light ratio , @xcite ) , whereas the virial mass is generally taken to be @xmath30 for the same region .
however , this can also be ascribed to the substantial cluster to cluster variations in the shape of the velocity dispersion profile ( @xcite ) due to the asphericity and substructure within individual clusters .
another indication comes from substructure , which is a complication in the analysis of clusters as individuals , but can be turned to advantage as an indirect , model dependent , @xmath21 estimator .
the results of substructure analysis remain controversial : some workers favor @xmath31 ( @xcite ) , while others suggest much smaller @xmath21 values ( @xcite ) .
the cnoc ( canadian network for observational cosmology ) cluster redshift survey is designed to derive a cluster @xmath26 within a homogenous , self - contained , sample where we can explicitly test for systematic errors .
the catalogues contain @xmath322600 redshifts in the fields of 16 clusters ( @xcite ) from which @xcite ) derived an average @xmath33 ( calculated with @xmath10 ) .
the clusters are consistent with having a universal @xmath0 value ( within the errors of about 25% ) independent of their velocity dispersion , mean color of their galaxies , blue galaxy content , or mean interior density .
the field galaxies in the dataset , with the same corrections , over the same redshift range , yield the closure mass - to - light ratio , @xmath34 , to be @xmath35 .
consequently the virial mass - to - light ratio implies @xmath36 at a mean redshift of 0.32 , where the error is the formal @xmath16 random error .
in @xcite the derived @xmath21 was erroneously corrected to a redshift zero @xmath26 .
the calculation uses a co - moving volume element ( @xmath37 dependent ) and hubble constant which are both redshift zero values which do not need further corrections .
testing the accuracy of the virial mass was a primary consideration in the design of the cnoc observations .
the data allow us to derive a radially resolved projected number density profile , @xmath38 , and a projected velocity dispersion , @xmath39 ( in the following , @xmath40 is used for projected radius ; whereas @xmath8 denotes the spherical co - ordinate ) .
the sample meets the two main requirements for deriving a radially resolved mass profile which are , first , accurate control of the background , and second , enough clusters to average over the aspherical complications of individual clusters . in section 2
we review the properties of our sample .
we define scaling velocities and radii which we use in section 3 to combine all but two `` binary '' clusters to create an `` ensemble cluster '' .
this averaging increases the signal - to - noise ratio of our measurements , and is sufficient to justify the assumption of an equilibrium , spherical , cluster in the analysis .
the clusters are sampled to densities in redshift space that are comparable to the field density , requiring a background subtraction which is discussed in section 4 .
the mean surface density profile of galaxies within the cluster is derived in section 5 , and fitted with a spatial density profile .
the projected velocity dispersion profile is derived in section 6 , and fitted to a radial velocity dispersion function for a variety of orbital shape assumptions . using these results
the mass - to - light profile is derived from jeans equation in section 7 . in section 8
we examine the relative colors and luminosities of field and cluster galaxies as a measurement of the differential evolution of the two populations .
the corrections and error budget are presented in section 9 , followed by our conclusions and discussion in section 10 .
the cnoc sample is designed to create a dataset that allows as complete internal control of all aspects of the cluster @xmath21 estimate as possible .
it is essential that the sample be useful for a test of the equilibrium hypothesis and whether the virial mass is biased in some way .
furthermore the sample is used to measure differential luminosity evolution between cluster and field galaxies . on the basis of some n - body simulation data ( @xcite )
it was argued that these could be best met within the constraints of available observational resources with a set of a dozen or so clusters at @xmath41 with a total of 3000 or so accurate redshifts .
the cluster sample was chosen from the einstein medium sensitivity survey catalogue of x - ray clusters ( @xcite ) to have a high x - ray luminosity , @xmath42 erg s@xmath43 , which helps guarantee that the clusters are reasonably virialized and have relatively high masses , making them easier to study .
the clusters chosen , table [ tab : r200 ] , are at moderate redshifts , @xmath44 , which has a number of advantages for mass estimation .
they are sufficiently distant that they have a significant redshift interval over which the density of foreground and background galaxies are nearly uniformly sampled in redshift . at @xmath14 , a typical cluster diameter of
@xmath45 ( comoving ) spans an angle of about 12 , which is sufficiently small that uniformity of photometry and sample selection is relatively easily assured .
this angle is also comparable to the field size of the canada - france - hawaii telescope ( cfht ) multiple - object - spectrograph ( @xcite ) , approximately 10 square .
observations were made at cfht in 24 assigned nights in 1993 and 1994 , of which 22 were usefully clear .
the sample and observational techniques are described in detail elsewhere ( @xcite ) .
the primary catalogues contain gunn @xmath8 magnitudes and @xmath46 colors for 25,000 objects , of which about 2600 have velocities , on the average accurate to about 100 .
about one - third of the sample is cluster galaxies .
the other two - thirds are field galaxies , although less than half of them are within the fair sample region defined by the band - limiting filter s passband and a @xmath47 buffer zone added to the upper and lower cluster redshift range . for reasons of observational efficiency , the survey is not `` complete '' in the usual sense .
hence , each object with a redshift has calculated weights which are the inverse of the magnitude selection function and the magnitude - dependent geometric selection function ( @xcite ) .
there are small changes in the cluster dynamical parameters of table [ tab : r200 ] from those in the global analysis ( @xcite ) .
the values here are derived using the final survey catalogues which accounts for the small changes from those given in earlier papers .
all calculations in this paper assume @xmath48 , @xmath49 and @xmath50 , although these choices do not affect our main conclusions in any significant way .
the primary goal of this paper is to measure any systematic biases in @xmath0 as an estimator of the field value .
this is done by independently measuring the mass and constraining differential luminosity evolution of cluster and field galaxies .
we adopt the virial mass , @xmath51 as the standard estimator , where 1 and @xmath52 are defined below in equations [ eq : sig ] and [ eq : rv ] .
the drawback of the virial mass is that it is a statistically `` inefficient '' estimator ( @xcite ) .
its positive features are that it is completely independent of the distribution of orbital shapes and it is not unduly sensitive to background contamination .
an alternate estimator is the projected mass ( @xcite ) , which we calculated but do not report .
its main drawback is that it is quite sensitive to background contamination of the data at large radii . for the standard velocity ellipsoid shape
the projected mass is always larger than the virial mass , which we find below already overestimates the cluster mass .
hence , we adopted the virial mass as our standard estimator .
the first step in testing the accuracy of @xmath11 is to construct a low - noise , effectively spherical , `` ensemble '' cluster , which averages over internal substructure and the variation of projected quantities with viewing angle .
our clusters span about a factor of two in velocity dispersion and hence are not a uniform set of physically identical objects . simply overlaying all the clusters onto a common center of mass in physical velocity and position units has some significant advantages ( an approach that will be pursued in another paper ) , but has the disadvantage that any gradients of density or velocity dispersion tend to be blurred out .
the natural scaling parameters are each cluster s rms velocity dispersion and the characteristic radius at which the cluster is expected to be in an effective equilibrium , which turns out to be a function of the rms velocity dispersion , alone , within our approximation .
deciding which galaxies in redshift space are cluster members is fundamentally ambiguous .
that is , a cluster galaxy with a completely reasonable line - of - sight velocity of @xmath53 appears in redshift space at 10 from the cluster s center of mass , far outside the virialized cluster and overlaying field galaxies .
this complication increases in severity at large distances from the cluster center , which our sample is specifically designed to probe .
one straightforward solution to this problem , which we adopt , is to explicitly subtract the mean density of field galaxies in redshift space from the cluster distribution .
the internal kinetic energy of the cluster is calculated from the characteristic velocity dispersion , @xmath54 .
we iterate the classical estimator @xmath55 where the @xmath56 are the peculiar velocities in the frame of the cluster and @xmath57 is the weighted mean redshift of the cluster .
the weights are calculated from the redshift and photometric catalogues to allow for statistical incompleteness of sampling ( @xcite ) .
the key to the use of this estimator is to have an objective choice of the galaxies that are likely to be cluster members .
the adopted method ( @xcite ) is as follows .
first a choice of the cluster redshift range is made from which a trial 1 is calculated .
then all galaxies between 5 and 15 1 are used to make a background estimate , which is subtracted from the weights of the galaxies selected to be in the cluster .
if the 1 calculated from the background subtracted weights is within the 68% bootstrap confidence range of the trial 1 ( @xcite ) , then the procedure is stopped ; otherwise , the redshift limits are increased or decreased to find a convergence . precisely the same galaxies as determined to be `` in the cluster '' from our 1 calculation
are given to the iterated bi - weight estimator ( @xcite ) .
the resulting estimates of the velocity dispersion are , on the average , @xmath58 times larger than from our method .
we find below that the global mean velocity dispersion of the combined dataset , normalized with our 1 values , is @xmath59 , whereas it should be unity , which is consistent with either scheme for calculating the velocity dispersions .
since the outcome of this investigation is that @xmath11 is biased upwards , the bi - weight calculation would lead to a further increase in the bias . for each cluster
a `` ringwise '' estimate of the virial radius of the observed distribution is calculated ( see @xcite for details ) : @xmath60 where @xmath61 and @xmath62 is the complete elliptic integral of the first kind in legendre s notation ( @xcite ) .
the angular extent of the sample is set by the observed field size .
the redshift range is as found in the velocity dispersion calculation .
the goal is that our data will include the entire virialized mass of the cluster .
analytic models ( @xcite ) and simulation data ( @xcite ) find that the virialized mass is generally contained inside the surface where the mean interior density is @xmath63 at the redshift of the cluster . the mean interior density within our measured @xmath52 is @xmath64 in terms of the measured dynamical parameters
this is @xmath65 where the hubble constant at @xmath66 is @xmath67 $ ] ( @xcite ) .
normally we take @xmath49 , @xmath68 and @xmath69 for an open frw model .
the mean density inside @xmath52 serves as a check as to whether the radial extent of the cluster is sufficiently sampled to determine a virial radius that includes most of the virialized mass .
table [ tab : r200 ] gives the values for @xmath52 , @xmath54 , and @xmath70 , in columns 3 , 4 and 5 , respectively .
the sampling over - density @xmath70 varies substantially from cluster to cluster , which mainly indicates the radial extent of our coverage of a given cluster .
the observations used here extend beyond 200 for most clusters .
the exceptions are ms1006@xmath7112 , ms1008@xmath7212 , and ms1455@xmath7122 where the data extend to about 2/3 to 3/4 of 200 .
the cluster s virial mass to k - corrected gunn @xmath8 luminosity , @xmath0 , and its errors are given in columns 6 and 7 of table [ tab : r200 ] .
the numbers are slightly different than those found in @xcite because these results are based on the finalized catalogues .
to calculate 200 , the radius where the mean interior density is @xmath73 , requires an assumption as to how the mass is distributed .
one possibility is that the mass is strongly concentrated to the center of the cluster such that all the mass measured by @xmath11 is inside the measured @xmath52 , in which case , @xmath74^{1/3}. \label{eq : rp200}\ ] ] this definition of @xmath75 is not particularly appropriate when @xmath76 , as it is for our best sampled clusters , because the cluster galaxies , and presumably the mass as well , extend beyond @xmath75 .
a superior 200 is calculated assuming a density - radius model which is extrapolated to the desired overdensity . simplicity , and a great deal of work on clusters of galaxies and the properties of dark halos in general
, suggests that the first approximation for the density - radius relation is the singular isothermal sphere , @xmath77 , in which case we define , @xmath78^{1/2},\ ] ] or equivalently , @xmath79 note that @xmath80 and @xmath81 .
that is , 200 is independent of the angular extent of the sampling , in so far as the cluster s 1 value has no radial dependence , which for the shallow gradients of the velocity dispersion profile found below is effectively true . at @xmath82 a low density , flat universe gives a value of 200 about 13% larger than in a open universe .
values of @xmath83 and 200 are given in columns 8 and 10 of table [ tab : r200 ] in units of physical for our cluster sample .
for our sample these two 200 estimates have very similar values , so the results are not sensitive to the mass model .
the jacknife error estimates for @xmath83 and 200 in are given in columns 9 and 11 , respectively .
all of the following analysis will use 200 as the scaling radius .
two of the clusters , ms0906 + 11 and ms1358 + 62 , are strong binaries , the first having an ill determined velocity dispersion and the second having a massive substructure on one side moving at 1000 relative to the cluster center .
these are dropped from all further use in this paper .
the scaled velocity - radius data for the remaining clusters are displayed in figure [ fig : cvt ] plotted using the brightest cluster galaxies ( bcgs ) as the projected center and the weighted average redshift as the velocity center .
all of the cluster properties presented in this paper are measured using galaxies with redshifts .
the clusters typically extend over a velocity range of @xmath84 , which in redshift space causes orbitting cluster galaxies to be intermingled with field galaxies within a 50 distance interval around the cluster .
background and foreground galaxies that are projected into the redshift space of the cluster must be statistically removed .
this survey was deliberately placed at moderate redshift to give a field sample over a considerable redshift column from which the background density can be accurately measured .
the background density is measured in the normalized velocity - radius space of figure [ fig : cvt ] .
it should be noted that the sampling in figure [ fig : cvt ] varies substantially with both velocity and radius which is addressed with redshift cutoffs and magnitude and position dependent weights ( @xcite ) .
the background and foreground are folded on top of one another within the unbiased redshift sample range of the band limiting filters which removes any linear gradients present in the redshift distribution .
there are two approaches to background subtraction used in the following calculations . for the surface density calculation we use the straightforward average over the sky area
the velocity dispersion calculation takes a more cautious approach to background subtraction .
a resolved background is derived in precisely the same range of projected radii for which @xmath85 is calculated .
the advantage of this approach is that the contamination of the cluster velocity profile with various nearby groups of galaxies is subtracted in the range of projected radii where they contribute to the background .
furthermore , the @xmath85 calculated from non - overlapping radial ranges are completely independent of each other .
the surface number density profile of the average cluster , @xmath38 , is the projection over velocity onto the radius axis , @xmath40 , of the data of figure [ fig : cvt ] . to find @xmath38 , we sum the statistical weights ( which correct for nonuniform sampling , @xcite ) for objects within 31 of zero velocity , divide by the bin area , and subtract the mean background .
we prefer the bcgs as the projected centers ( @xcite ) , but repeat most of the analysis using the peak of the smoothed density profile as the centers , which gives a useful view of systematic effects in the analysis .
the mean background volume density in velocity space , @xmath86 , is calculated by summing over the entire range of @xmath87 using the background in the velocity range of 5 to 25 1 .
we find that @xmath88 , in units of the normalized total numbers per 1 per @xmath89 .
figure [ fig : back ] shows the variation of the average background density with velocity separation ( in normalized units ) from the cluster . between 3 and 61 , the background calculated from the full sample ( left panel ) is higher than expected for a high velocity tail like a gaussian .
this excess is statistically significant at about the 3 standard deviation level , and results from various small groups that happen to be behind the low redshift clusters .
the right panel of figure [ fig : back ] shows the background calculated after the 7 clusters with @xmath90 are dropped .
the rapid decline of clustering with redshift leads to a smoother distribution in redshift in the right panel .
the mean background is calculated over the entire 5 to 251 interval for the full sample , so a local excess makes little difference . as a sidelight we note that near the cluster center , 1 - 31 , the velocity distribution is possibly somewhat `` squarer '' than a gaussian distribution .
that is , for a gaussian the ratio of the numbers between 1 and 21 and 0 to 11 is about 1/3 , whereas we have more than 1/2 .
this is likely an indicator that the cluster velocity ellipsoids are somewhat radially anisotropic ( @xcite ) . to check the sensitivity of the derived @xmath38 to the assumed background level we repeat the calculation using values for @xmath86 of 0.0048 and 0.0082 , which are 2 standard deviations from the mean value .
the only part of the profile substantially affected is beyond 200 , with alternate values being just within the 1 standard deviation range of the estimated density at that point .
that is , for our adopted bin widths the statistical error due to sample size generally exceeds the uncertainty in the background subtraction .
the fitted profiles described below have scale radii altered by 1.5% for a 26% variation of the background .
we conclude that the background subtraction is not a dominant source of error for @xmath38 .
the background subtracted @xmath38 are displayed in figure [ fig : surf ] and tabulated in table [ tab : surf ] .
the vertical normalization of @xmath38 is a fixed , but arbitrary , value . in table
[ tab : surf ] the number density @xmath91 ( column 2 ) is measured about the bcg , and @xmath92 ( column 4 ) is measured about the peak of the 45 gaussian smoothed density map . within @xmath93 of the center
the density profile is strongly dependent on the choice of center . because there is usually no galaxy at the point of peak density
, there is a slight dip in the central surface density when measured about the peak of the smoothed density distribution .
the choice of center makes little difference to the measured @xmath38 at radii beyond @xmath93 , approximately 100 , but it does affect fits to a specific model .
the point of peak density generally has sufficiently large errors that it could be consistent with being at the bcg , hence we always prefer the bcg center for physical reasons ( @xcite ) . outside of @xmath6
the surface density is comparable to the background but the data are very sparse , which together cause the surface density at large radius to be quite uncertain .
no data beyond @xmath6 are used in the velocity dispersion analysis .
the surface luminosity density profile was also derived ; but , being weighted to the brightest galaxies , it is a noisier quantity than the number density .
we find that the luminosity density profile shape is essentially the same as the density distribution . the spatial galaxy number density as a function of radius , @xmath94 , is needed to compare the galaxy number profile with the cluster mass profile .
the relation between the volume density and the surface density is , @xmath95 a statistically adequate , analytically convenient , model that describes these data is the hernquist ( alternatively designated as an @xmath96 ) model ( @xcite ) .
we fit the data centered on the bcg of figure [ fig : surf ] to the projection of the volume density model , @xmath97 this procedure for finding a @xmath94 consistent with the @xmath38 data is not unique , there being both the minor variation of using one of several other entirely reasonable fitting functions , or , adopting a completely different approach such as the non - parametric , maximum likelihood method ( @xcite ) modified to allow for background subtraction .
the function @xmath98 ( @xcite ) is also a statistically acceptable fit with a slightly increased @xmath99 .
our measured @xmath38 falls faster than this function predicts at large @xmath40 , but at these radii our measurement is relatively uncertain .
fitting the surface density data with equation [ eq : nu ] finds that @xmath100 for the clusters centered on their bcgs .
the reduced chi - squared statistic , @xmath101 , is 0.74 for @xmath102 degrees of freedom , which is an acceptable fit .
using the @xmath38 measured about the peak light gives a considerably larger scale radius , @xmath103 , with @xmath104 .
this larger value of @xmath105 is at least in part due to a blurring of the density profile by statistical errors in measuring the position of peak number density .
although we believe it is a less appropriate choice for the cluster centers for these data , we carry it along in the analysis as an indicator of systematic errors .
the measurement of @xmath85 is a second moment of the velocity distribution about the local mean velocity at projected radius @xmath40 , although the choice of central moment makes essentially no difference to the result .
the projected velocity dispersion at @xmath40 , @xmath85 , is calculated using an iterative clipping of the high velocity dispersion tails with a technique similar to that used for the rms velocity dispersion in @xcite , where @xmath54 is calculated from the background subtracted data within @xmath106 of zero velocity , where @xmath107 .
the iteration begins with @xmath108 in the 1 normalized velocities . for the next four iterations the @xmath85 from the previous iteration
is used for clipping . for a 10% change in the clipping level , @xmath109 , the resulting velocity dispersion changes about 1% at large radii .
the error flags are calculated from a full bootstrap resampling of both the cluster and field data . rather than radially binning the data
, we use a moving average , which includes all galaxies , both cluster and field .
table [ tab : sig ] gives the results with a 51 point moving average which are plotted in figure [ fig : sigfit ] at intervals of 0.1 in 200 .
these points are always independent for 51 point averaging .
averages using 31 to 201 points give consistent results , the 31 point result being quite noisy , and the 201 point beginning to smooth the velocity dispersion gradient .
the results of a 101 point moving average are shown in figure [ fig : sig ] . for 101 point averaging the error flags
are not independent outside of @xmath110 .
the random errors of the derived @xmath85 of the ensemble are dominated by cluster to cluster fluctuations . within an individual cluster ( for instance a2390 )
the random errors are substantially smaller , but there are large differences in velocity dispersion profiles from cluster to cluster ( @xcite ) which we attribute to projection effects and substructure . at large radii ,
dynamical measurements from a single cluster likely do not give reliable enclosed masses because of the complications of infall , triaxiality and substructure .
the idea of creating the ensemble cluster is of course to diminish these systematic variations to the level where an effectively spherically symmetric average profile emerges . to derive a @xmath111 which is consistent with the observed @xmath85 we use the same approach as we used for the density profile
that is , we choose a reasonable functional form , then adjust 2 parameters , a scale length and a normalization , until its calculated projection , @xmath112 minimizes the @xmath99 . in this equation @xmath113 is the velocity anisotropy parameter . for the purposes of this paper @xmath114
will be taken to be a constant , although we do use a range of values .
the adopted @xmath111 function should be finite at the origin , at large @xmath8 it should tend to a keplerian @xmath115 for a convergent mass distribution , and it should be a simple smooth function in between .
it is important that this function not assume that the galaxy populations are self - consistent with the mass density of the potential .
we adopt the simple form , @xmath116 for the radial velocity dispersion .
this function , with b=@xmath117 and assuming @xmath118 , is the solution of the jeans equation ( equation [ eq : jeans ] below ) for @xmath119 and @xmath120 in our assumed @xmath94 .
the scale lengths @xmath121 and @xmath105 are fitted separately , usually finding that @xmath121 is larger than @xmath105 .
the nonlinear fitting procedure requires that the errors be symmetric about the observed value , so we take the larger of the upper and lower error range at each point .
the fit is done assuming @xmath122 and @xmath123 .
the projected fit and the 51 point data are shown in figure [ fig : sigfit ] .
the @xmath101 of the two parameter fit , an amplitude and length scale , is @xmath124 for all @xmath114 values .
to compare the relative distribution of mass and light we integrate our fitted @xmath94 to give @xmath125 , the `` light profile '' ( actually a number density profile ) , which is converted to a mass profile using a normalized global @xmath0 , calculated in the same manner as for the individual clusters .
then we compare the mass predicted by the light - traces - mass assumption to the @xmath126 derived from the fitted @xmath111 and @xmath94 using jeans equation , @xmath127 .
\label{eq : jeans}\ ] ] we will usually refer to the mass derived from this equation as @xmath128 , the stellar hydrodynamical mass . it must be borne in mind that equation [ eq : jeans ] is a moment of the collisionless boltzmann equation , and will not necessarily give a physical result for all @xmath114 , being particularly susceptible to failure for very radial orbits ( @xcite ) .
in addition , radial models are usually dynamically unstable to the radial orbit instability ( @xcite ) . however , the galaxies are only tracers , so in principle they could be on much more radial orbits than the underlying dark matter . on the other hand , it seems likely that galaxies orbiting past the center may be tidally destroyed or simply well mixed so that the galaxy velocity ellipsoid near the center may become less radial than the dark matter .
equation [ eq : jeans ] is integrated with the boundary condition that @xmath129 as @xmath130 .
alternative boundary conditions are that the velocity dispersion goes to zero at the turnaround radius ( @xcite ) , or , that there is an outer region of nearly radial infall which makes a transition to virialized motion ( @xcite ) .
n - body simulations ( @xcite ) show that on the average , the approximation that the virialized motion continues to 200 is a good one .
for constant @xmath114 equation [ eq : jeans ] has a formal solution , @xmath131 satisfying the jeans equation is a necessary condition for an equilibrium to exist , but it is not sufficient and is certainly not a guarantee that a spherical distribution is dynamically stable .
models with a constant @xmath132 are unphysical , because the implied mass is negative in the inner regions ( very radial orbits lead to @xmath133 , not the @xmath134 here ) .
this , along with the results of numerical simulations , leads to our preference for @xmath135 .
the model lines plotted on figure [ fig : sig ] are for @xmath136 and 1 , assuming @xmath137 in eq .
[ eq : solution ] , using the fitted @xmath94 from eq .
[ eq : nu ] , and normalizing the projected rms velocity dispersion over the observed radial range to the observed value .
the numerical complication in deriving @xmath126 from equation [ eq : jeans ] is that the logarithmic gradients @xmath138 and @xmath139 must be calculated . numerically evaluating these gradients from the data requires some form of smoothing to be stable , which our fits to smooth functions effectively accomplish .
we will carry the errors of the projected data through the analysis to help give an assessment of the result .
the radial dependence of the shape of the velocity ellipsoid , as defined by the value of @xmath114 , is not empirically known for this sample , although this does not introduce the same uncertainty as it does for finding central dark masses . as
the velocity ellipsoid becomes more and more radial at large radii , the mass implied for a measured increases .
the inferred mass has nearly a full order of magnitude uncertainty when a velocity dispersion is available at only a single point with no knowledge of the shape of the velocity ellipsoid ( @xcite ) .
because we have measured the velocity dispersion over a wide radial range the uncertainty due to velocity dispersion anisotropy becomes quite small , as shown below .
n - body simulations of objects that form via collapse find that @xmath140 ( @xcite ) , although it is clear now that `` vacuum '' boundary conditions give far more velocity anisotropy than is realistic in a cosmological setting ( @xcite ) where substantial external torques provide angular momentum to large radius orbits .
furthermore , strongly radial models are unstable ( @xcite ) , although galaxies , being nearly massless within a cluster , could have a different velocity ellipsoid than the dark matter .
for the nearly power law hierarchical clustering that characterizes the growth of clusters in model universes one expects that the velocity ellipsoid will have only a weak radial dependence , because cluster buildup is self - similar .
this is borne out in n - body simulations ( @xcite ) which find @xmath141 over the radial range of interest for a wide range of @xmath26 and clustering models .
the depletion of hi in coma spirals suggests ( @xcite ) that their velocity ellipsoid is not strongly radial , and is consistent with being isotropic . a test of the accuracy of the virial mass as an estimator of the total mass virialized in the cluster is whether the product of the virial mass - to - light ratio computed from our average cluster and the integrated interior luminosity , @xmath125 , is equal to the stellar hydrodynamical mass , @xmath128 , derived from eq .
[ eq : jeans ] .
that is , we derive the virial mass bias , @xmath142 , from @xmath143 where @xmath125 is simply the volume integral of @xmath94 .
the normalization of @xmath125 , @xmath144 , is arbitrary since it cancels in eq .
[ eq : mlbias ] .
the normalizing virial mass , @xmath145 , is derived for the ensemble cluster in the same way as for the individual clusters .
two aspects of @xmath142 are of primary interest .
first , @xmath146 should be unity if the @xmath147 calculated from the average cluster is an unbiased estimator for all material inside 200 , the radius expected to contain most of the virialized mass .
second , the radial gradient of @xmath142 measures to what degree light traces mass in the average cluster .
the first issue is the one of most relevance for the problem of @xmath21 estimation .
it should be noted that whatever the value of @xmath142 turns out to be here , it is a quantity that is meant to give the systematic error in measuring cluster mass using red selected galaxies in rich , reasonably relaxed clusters .
it would be inappropriate to apply the same factor to merging clusters or a galaxy sample selected to emphasize high star formation rates , for example .
the galaxy number density profile , which we will call the light profile , is normalized to a mass using a mass - to - light ratio calculated in the same manner as done for our clusters individually .
there are two complications in this calculation which need to be examined .
first , the clusters are not all sampled to the same fraction of 200 .
second , both the @xmath52 and the total light are calculated without any allowance for background contamination . to investigate these effects we calculate @xmath147 using the mean profile , both with and without background subtraction , limiting the data at various @xmath148 .
the results are shown in figure [ fig : mlrad ] . remarkably , the ratio @xmath147 is almost entirely independent of background contamination and sampling radius , once @xmath149 .
the lower line on the figure is made using the surface density without background subtraction , which causes @xmath150 , hence @xmath145 , to be overestimated , but the summed luminosity is also overestimated . the pleasing result for this sample
is that the two overestimates almost exactly cancel out so that @xmath147 calculated in the usual manner has no sensitivity to background .
furthermore , @xmath147 has very little dependence on the sky coverage of the cluster , provided that the data extend beyond @xmath151 .
figure [ fig : mlrad ] shows that the directly calculated average @xmath147 is high in the inner regions , but then drops to a very stable value at large radius .
the normalizing virial mass is @xmath152 , where @xmath153 and the parameters are the measured @xmath154 and @xmath150 for the combined cluster data .
we use the quantities for the full sample , with background subtraction , although this choice makes little difference to the results . with this normalization
, @xmath146 should be unity if @xmath11 is unbiased . the internal @xmath0 bias , @xmath142 , from eq .
[ eq : mlbias ] is displayed as a function of radius in figure [ fig : mol ] for @xmath122 and @xmath123 .
the bias would be unity everywhere if the product of @xmath147 and @xmath125 was equal to the stellar hydrodynamical mass . note that if @xmath155 the implied mass is negative at small radius .
the primary goal of this paper is to correct the virial mass - to - light ratio to the field @xmath1 value .
the average virial mass bias for the bulk of the virialized system estimates the systematic offset in the virial mass , and is reasonably estimated as the value of @xmath156 evaluated at 200 , displayed in figure [ fig : mvb ] as a function of the assumed @xmath114 .
because the gradient is small the exact radius used does not make a big difference .
the error in @xmath146 is calculated from the error in the fit to the @xmath157 coefficient in @xmath111 , the error in the other quantities being negligible in comparison .
the 3 symbols are for absolute magnitude cutoffs of @xmath158 - 18.5 , -19.0 , and -19.5 mag , respectively , of the entire calculation .
the points offset in @xmath114 , with the dashed error flags , use the peak density as the centers and are always poorer fits , hence have larger errors .
these points are included to illustrate the small systematic differences .
all the results are statistically identical .
we find @xmath11 , as normally calculated , is always an overestimate of the mass , whose value is weakly dependent on the assumed @xmath114 value . for this sample
the bias is minimal ( @xmath156 closest to unity ) near @xmath159 for which it is equal to @xmath160 .
all @xmath114 give statistically consistent bias values , although there is clear trend of decreasing @xmath156 with increasing @xmath114 .
the inferred radial gradients of @xmath156 for different @xmath114 are shown in figure [ fig : mvg ] .
the gradient is calculated from the slope of the lines in figure [ fig : mol ] . only for very radial orbits , @xmath161 ,
is there even a mildly significant rising mass - to - light ratio with radius , and those are the models with negative central mass .
the gradients are sufficiently small that for a wide range of @xmath114 they are consistent with there being no gradient of @xmath1 . in no case
are the gradients sufficiently large that there is any possibility that the cluster has a dark matter halo that is significantly more extended than the visible galaxies . we have found that the virial mass , as we have calculated from the data , is about a 20% overestimate of the true mass .
the origin of this factor needs to be quantitatively understood .
we believe that this bias is simply the approximation that the surface pressure term in the virial equation is zero . for our fitted functional forms for the density profile and the velocity dispersion profile , ( equations [ eq : nu ] and [ eq : sigfit ] , respectively ) , we can provide an estimate of the expected `` truncation bias '' .
truncating a density profile of the form eq .
[ eq : nu ] , @xmath162 , will cause the true mass contained within that radius to be overestimated , as quantified below .
the potential energy scalar , @xmath163 , of the density distribution truncated at @xmath164 is , @xmath165 calculating the virial radius as @xmath166 we find that @xmath167 the estimated virial radius is @xmath168 if the distribution is sampled to infinity distribution , whereas for our combined clusters we have measured the virial radius to be @xmath169 ( using the @xmath105 found from the fit of the distribution with background subtraction ) .
this is consistent , within the errors , with our truncation radius @xmath170 and eq .
[ eq : rg ] for the combined sample .
the clusters as individuals all have truncation radii less than or equal to this .
if the galaxy distribution follows the mass density the mass contained within @xmath164 is estimated from the `` classical '' virial theorem relating the kinetic energy , @xmath171 , @xmath163 , and the surface term , @xmath172 , through @xmath173 , as @xmath174 .
the usual assumption for dynamical systems is that @xmath172 is evaluated as @xmath175 , where it is zero .
this is not correct in detail for clusters of galaxies , where this term is a significant correction . for the particular case of @xmath119 ,
it was noted above that the solution of jeans equation is @xmath176 $ ] .
hence the truncated kinetic energy , @xmath171 , is , @xmath177 for the case of the mass being distributed like the galaxies , the ratio of the estimated mass contained within @xmath164 to the true mass contained is , @xmath178 note that this expression could also be derived from @xmath179 , as discussed in @xcite . at any finite radius
this estimated mass is greater than the true global value . at @xmath180 , @xmath181 , @xmath182 and @xmath183 it evaluates to 8/5 , 5/4 , 8/7 and 35/32 , respectively .
that is , the mass estimated from the averaged distribution is expected to overestimate the true value by about 10 to 20% .
clusters that are not sampled to large radii will have a larger truncation bias but smaller background contamination in the projected quantities .
the data in most of our clusters extend to 1 to 2 200 , roughly consistent with the measured @xmath156 .
a crucial step in the derivation of @xmath21 from cluster @xmath1 values is to correct for any differential luminosity evolution between the cluster and field galaxies .
the luminosities of galaxies per unit total mass can , in principle , either increase or decrease depending on what happens when galaxies fall into a cluster .
the empirical evidence is that once galaxies have spent some time in rich , high x - ray luminosity clusters like those in our sample , star formation largely ceases .
however , there are at least two routes to this end point for a gas rich disk galaxy .
galaxies entering the cluster could have their gas largely stripped away , or , a burst of star formation upon cluster entry could boost luminosity prior to depleting the gas . in either of these cases , lowering the star formation rate leads to a decrease in a galaxy s luminosity by an amount that depends on its entire star formation history .
galaxy merging always decreases numbers , but can lead to luminosity increases if accompanied by star formation .
all these possibilities are testable in one way or another , and some substantial constraints on differential evolution can be imposed using the data at hand .
it is well known that the fraction of cluster galaxies with blue colors is a strongly increasing function of redshift , the butcher - oemler effect ( @xcite ) .
it should be noted that cluster galaxies are , on the average , never bluer than field galaxies , implying that , on the average , star formation always declines into a cluster .
if the infall of field galaxies into clusters induces a temporary increase star formation and hence increases cluster galaxy luminosities relative to the field , then it should be apparent as a change in the color distribution ( and the strength of the [ oii ] lines ) .
this is not seen in our data ( @xcite ) .
on the basis of these considerations we expect that cluster galaxies are of intrinsically lower luminosities than the field galaxies .
our sample of galaxies extends from the cluster center to the distant field , allowing population gradients to be examined as a function of a redshift space radial variable , @xmath184 .
a convenient definition of such a variable is based on our normalized co - ordinates , where velocity differences are scaled to 1 and projected radial co - ordinates are scaled to 200 , so that the dimensionless redshift space separation from the center is @xmath185 , where @xmath40 is the projected radius from the center and @xmath186 is the rest - frame velocity difference from the average redshift .
note that at large @xmath184 the galaxies come exclusively from the redshift direction .
figure [ fig : czr ] shows the color - radius relation of the full sample .
all galaxies with k - corrected absolute @xmath8 magnitude , @xmath187 , brighter than @xmath188 mag are plotted .
the colors are corrected to an empirically normalized redshift - independent color , @xmath189^{-1 } \label{eq : gr}\ ] ] ( @xcite ) .
we note that there is no color gradient in the field , but one appears at @xmath190 , where we expect galaxies to enter the virialized region of the cluster .
there is no evidence in the colors for excess star formation at the edge of the cluster ; indeed , there are no cluster galaxies systematically bluer than field galaxies . hence , the distribution of galaxy color with radius ( fig .
[ fig : czr ] ) is consistent with infalling galaxies terminating their star formation with little or no `` starbursting '' once they enter the cluster , as detailed analysis of the cnoc data for a2390 has demonstrated ( @xcite ) . a sample extending to a larger projected radius is needed to directly demonstrate kinematic evidence for infall of galaxies into the clusters .
however , given the gravitational mass of the cluster , the turnaround radius is straightforwardly derived ( @xcite ) such that field galaxies must be falling into the cluster out to the turnaround radius , approximately @xmath191 for the model profile here . to a significant degree
the cluster must be composed of former field galaxies .
in such an infall scenario the cluster galaxies statistically share the star formation history of the field , prior to cluster entry . the only gross difference between clusters and field
is that clusters contain an unusual population of luminous central galaxies .
the bcgs only make a difference at small @xmath184 , and generally contribute well less than 10% of the light . in spite of the substantial radial color gradient of figure [ fig : czr ]
there is very little change in the @xmath8 band luminosity distribution of figure [ fig : lzr ] .
measurements of the surface brightness evolution of both disks and bulges arrives at the conclusion that the amount of differential evolution between the field and the cluster luminosities is relatively small ( @xcite ) .
there is no significant difference in the field galaxy luminosities or colors at any radius beyond @xmath6 , as shown in figure [ fig : lzr ] .
that is , the field galaxies near the cluster appear to be indistinguishable from those far away .
a generic prediction of `` natural bias '' models of cluster galaxy formation is that the galaxies that eventually fall into the cluster should form earlier because the overdensity of the incipient cluster speeds up the local time scale relative to the mean universe .
possibly this leads to a higher rate of conversion of gas to stars , causing cluster galaxies to be more luminous than their counterparts in the field .
the color gradient appears at about @xmath6 , approximately where one expects cluster x - ray gas to be encountered .
any fading or brightening of galaxies as they fall into the cluster can not be large . without the bcgs we measure a mean @xmath187 in the average cluster , @xmath192 , which is statistically identical to the field .
however , this averages together blue and red galaxies , which contribute differently to the field and cluster , masking a real fading . splitting at @xmath193 ( eq . [ eq : gr ] ) finds that both the blue and red galaxies show a decline in their mean brightness , above our limit , of @xmath194 mag ( @xmath195 for red , @xmath196 for blue ) between field and cluster . splitting the sample into a low and high redshift sample shows that the fading between red galaxies in cluster and field increases to about 0.2 mag at low redshift .
a full luminosity function fit ( h. lin , private communication ) finds a 0.2 mag field to cluster fading , but with large errors correlated with the faint end slope . in any case
there is no evidence within these data for an excess in stellar luminosity per unit dark matter in cluster galaxies over field galaxies .
the corrected @xmath8 band mass - to - light ratio in the field and these clusters does not vary much simply because there is so much `` old '' red light which is not altered during cluster infall .
the @xmath26 estimate ( @xcite ) from cluster virial masses and the conversion of the cluster luminosity ( or numbers of galaxies ) to an equivalent co - moving volume in the field depends critically on the accuracy of the virial mass and the average luminosity of field and cluster being identical or having a measurable differential evolution .
the expression for @xmath21 that we used as our best estimator is @xmath197 .
the field luminosity was estimated to be @xmath198 where the errors were determined from a jacknife analysis .
the value and random errors of @xmath199 were estimated as @xmath200 . from these results we concluded that the virial mass estimator gives @xmath36 .
however , this result needs to be corrected for errors in @xmath0 relative to the field .
we find that the virial mass needs to be multiplied by @xmath201 to give the shd mass at 200 .
the correction is a consequence of measuring the virial mass from a truncated distribution and using data beyond 200 .
the values of @xmath156 for @xmath202 are all within the errors of this value .
we also find a significant mean luminosity differential between field and cluster which requires that the cluster luminosity be boosted by @xmath7 mag to give the field value .
together these corrections reduce @xmath0 to a field @xmath203 and hence @xmath204 .
the errors in the various quantities are given in table [ tab : err ] , along with their quadrature and linear sum .
the main errors in our result are simply the statistics of the luminosity density and the average cluster @xmath0 values .
the systematic errors are estimated from the extreme versions of the analysis that we find for alternative @xmath114 values , choice of center , and color and redshift samples of the galaxies . to obtain an @xmath26 value accurate to about 10%
would require a sample approximately four times larger .
the total random error , 30% , is estimated from the quadrature sum of the individual errors
. some of the component errors could be partially correlated which would increase the total error somewhat .
the worst case possibility is if all the errors are completely correlated , in which case the error is the linear sum of the errors , 73% .
this must be a substantial overestimate because many of the errors come from completely uncorrelated sources .
the systematic errors in table [ tab : err ] are harder to estimate .
other choices of density centers and @xmath114 values give answers that allow error estimation for these parameter variations .
the least satisfactory situation is the estimate of fading between cluster and field .
our observations contain no evidence for excess star formation between cluster and field , so we feel that an increase in the average light can be ruled out .
similarly , fading of much more than 0.5 mag in our @xmath8 band light would lead to such gross differences between cluster and field that it would be immediately visible .
furthermore , for such a large fading to occur would likely require a stellar population weighted towards short lived luminous stars . to constrain further the differential luminosity evolution requires more detailed observations of the individual galaxies .
the overall goal of the cnoc survey is to use clusters of galaxies to derive a value of @xmath26 with a well determined error budget with particular emphasis on eliminating systematic errors .
the major innovations of our analysis are that it is completely self - contained , with most assumptions being testable , and that the error estimates are derived from the data themselves .
the dominant source of error is random cluster - to - cluster variations due to projection and substructure , rather than the internal error from individual clusters .
the global mass - to - light ratio ( in our photometric system ) of our sample clusters is constant within our typical errors of 20 - 30% at a value of @xmath200 . over the same redshift range we measure the closure value , @xmath34 to be @xmath205 ( @xcite ) .
we have shown that at 200 that the ratio of the dynamical mass to the cluster light , @xmath206 , with essentially no dependence on assumptions about velocity anisotropy . after allowing for the fading of cluster galaxies relative to the field , @xmath7 mag , and the bias in estimating @xmath11
, we estimate that @xmath1 field is @xmath207 , implying @xmath204 .
the objectively evaluated random errors and estimated systematic errors are given in table [ tab : err ] .
a strength of these data is that they cover the entire virialized cluster , extending out to where the overdensities are low , about @xmath208 , and hence reach to sufficiently large radii that the cluster @xmath1 should represent the @xmath1 of the infalling material .
if segregation of dark matter mass and galaxies somehow occurs during infall , prior to to virialization in the cluster , then that could produce an artificially low @xmath1 .
we find that the galaxy population outside the cluster has no measurable gradients .
a benefit of this redshift range is that the galaxy populations in clusters more strongly resemble the field than they do at low redshift , tightening the @xmath1 argument .
our @xmath26 estimate is similar to the least action principle result of @xmath209 ( @xcite ) which probes larger scales which have much smaller density contrast .
our result is inconsistent with an @xmath26 of unity in any `` cold '' matter ( less than about 1000 ) that falls into clusters .
furthermore the result is well below the @xmath210 interpretation of the large scale streaming velocity results ( @xcite ) . in particular
our result in combination with the cosmic virial theorem estimates of @xmath211 ( @xcite ) implying that these galaxies are relatively unbiased mass tracers .
because clusters assemble their mass from regions approximately 20 across , the difference between the streaming results and our determination is substantial .
that is , there is little room for extra matter which supports density perturbations on larger scales but does not enter clusters over our range of observed redshifts .
the @xmath26 we derive is a reasonably comfortable 3 standard deviations above the upper end of the fraction universe in baryons , @xmath212 , ( @xcite ) , as the x - ray mass measurements also indicate ( @xcite ) .
the x - ray analyses for the ratio of the x - ray gas mass , @xmath213 , to the total mass , @xmath214 , find @xmath215 ( approximately adjusted to @xmath216 , but see @xcite )
. in combination with our results for @xmath26 this line of reasoning gives @xmath217 with an error of about 35% .
consequently cluster baryons are consistent with other bbn indicators for @xmath218 , although with a fairly large error range . the fact that the galaxy numbers and the cluster mass are distributed in an identical manner rules out any significant velocity bias between the cluster galaxies and the dark matter .
that is , a small ( say 20% ) decrease in the velocity dispersion of cluster galaxies relative to the dark matter would lead to a substantial segregation of the cluster galaxies relative to the cluster mass ( @xcite ) .
the conclusion to be drawn is that the galaxy tracers selected in @xmath27 n - body simulations do not form like the galaxies in the real universe .
the implications of this are unclear .
it is possible to identify relatively unbiased galaxy tracers in n - body simulations ( @xcite ) so the problem may be within the simulations .
a more interesting possibility is that the understanding of galaxy formation and mass clustering that is incorporated into the simulations has some substantial deficiencies .
there are several improvements in this analysis that could be made with more data , mainly at large radii .
more data at large radii would reduce the variance resulting from substructure and sample all the clusters to similar 200 , diminishing the complications of the @xmath199 bias correction . with data extending beyond
@xmath219 one enters the infall regime and expects to see the `` compression effect '' in the redshift space density contours of the ensemble cluster ( @xcite ) , which would empirically demonstrate that infall is occurring , and would give a measurement of @xmath220 in the nearly linear regime . with many more velocities in the cluster bodies
the shape of the velocity ellipsoid can be directly measured , rather than taken from n - body simulations as was done here .
we thank the many members of cnoc who helped us obtain and reduce these data .
comments from the referee , doug richstone , lead to substantial improvements in the presentation .
we thank cfht for the technical support which made these observations feasible .
nserc provided financial support for rgc and hkcy .
abell , g. o. 1965 , , 3 , 1 abraham , r. g. , smecker - hane , t. a. , hutchings , j. b. , carlberg , r. g. , yee , h. k. c. , ellingson , e. , morris , s. , oke , j. b. , davidge , t. 1996 , , in press bahcall , j. & tremaine , s. d. 1981 , , 244 , 805 barnes , j. , goodman , j. , & hut , p. 1986
, , 300 , 112 bean , a. j. , efstathiou , g. , ellis , r. s. , peterson , b. a. & shanks , t. 1983 , , 205 , 605 beers , t. c. , flynn , k. & gebhardt , k. 1990 , , 100 , 32 bertschinger , e. 1985 , , 58 , 39 binney , j. & tremaine , s. 1987 , _ galactic dynamics _ , ( princeton university press ) bird , c. m. 1994 , , 107 , 1637 butcher , h. & oemler , a. 1984 , , 285 , 423 carlberg , r. g. & dubinski , j. 1991 , , 369 , 13 carlberg , r. g. 1994 , , 433 , 468 carlberg , r. g. , yee , h. k. c. , ellingson , e. , pritchet c. , abraham , r. , smecker - hane , t. , bond , j. r. , couchman , h. m. p. , crabtree , d. , crampton , d. , davidge , t. , durand , d. , eales , s. , hartwick , f. d. a. , hesser , j. e. , hutchings , j. b. , kaiser , n. , mendes de oliveira , c. , myers , s. t. , oke , j. b. , rigler , m. a. , schade , d. , & west , m. 1994 , , 88 , 39 carlberg , r. g. , yee , h. k. c. , ellingson , e. , abraham , r. , gravel , p. , morris , s. m , & pritchet , c. j. 1996a ( c96 ) , , 462 , 32 cole , s. & lacey , c. 1996 , , 281 , 716 copi , c. j. , schramm , d. n. , & turner , m. s. 1995 , science , 267 , 192 crone , m. m. , evrard , a. e. , & richstone , d. o. 1994 , , 434 , 402 davis , m. & peebles , p. j. e. 1983 , , 267 , 465 dekel , a. 1994 , , 32 , 371 dekel , a. , bertschinger , e. , yahil , a. , strauss , m. a. , davis , m. , & huchra , j. p. 1993
, , 412 , 1 dubinski , j. 1993 , , 401 , 441 efron , b. & tibshirani , r. 1986 , _ statistical science _
, 1 , 54 efstathiou , g. , ellis , r. s. , & peterson , b. a. 1988 , , 232 , 431 fillmore , j. a. & goldreich , p. 1984
, , 281 , 1 fisher , k. b. , davis , m. , strauss , m. a. , yahil , a. , & huchra , j. p. 1994
, , 267 , 927 gioia , i. m. & luppino , g. a. 1994 , , 94 , 583 gioia , i. m. , maccacaro , t. , schild , r. e. , wolter , stocke , j. t. , morris , s. l. , & henry , j. p. 1990
, , 72 , 567 gott , j. r. & gunn , j. 1972 , , 176 , 1 gott , j. r. & turner , e. l 1976 , , 209 , 1 gunn , j. e. 1978 , in _
observational cosmology _ , eds .
maeder , a. , martinet , l , & tammann , g. 1978 ( geneva observatory : sauverny ) den hartog , r. 1995 , _ the dynamics of rich galaxy clusters _ , ph .
d. thesis , leiden university henry , j. p. , gioia , i. m. , maccacaro , t. , morris , s. l. , stocke , j. t. , & wolter , a. 1992 , , 386 , 408 hernquist , l. 1990 , , 356 , 359 kaiser , n. 1987 , , 227 , 1 van kampen , e. 1995 , , 273 , 295 katz , n. & white , s. d. m. 1993 , , 412 , 455 kent , s. & gunn , j. e. 1982 , , 87 , 945 lefvre , o. , crampton , d. , felenbok , p. , & monnet , g. 1994 , , 282 , 340 limber , d. n. 1959 , , 130 , 414 loveday , j. , efstathiou , g. , peterson , b. a. , maddox , s. j. 1993 , , 390 , 338 merritt , d. 1984 , , 313 , 121 merritt , d. & tremblay , b. 1994 , , 108 , 514 mohr , j. j. , evrard , a. e. , frabicant , d. g. , & geller , m. j. 1995 , , 447 , 8 navarro , j. f. , frenk , c. s. , & white , s. d. m. 1995 , apj submitted oort , j. h. 1958 , in _ la structure et lvolution de lunivers _ , onzime conseil de physique , ed .
r. stoops ( solvay institute : bruxelles ) p. 163 palmer , p. l. & papaloizou , j. 1987 , , 224 , 1043 peebles , p. j. e. 1970 , , 75 , 13 peebles , p. j. e. 1993 , _ principles of physical cosmology _
( princeton university press : princeton ) press , w. h. , teukolsky , s. a. , vetterling , w. t. , & flannery , b. p. 1992 , _ numerical recipes in c _ ( cambridge university press ) pryor , c. & geller , m. j. 1984 , , 278 , 457 ramella , m. , geller , m. j. , & huchra , j. p. 1989
, , 344 , 57 regs , a. & geller , m. j. 1989 , , 98 , 755 richstone , d. o. , loeb , a. , & turner , e. l. 1992 , , 393 , 477 richstone , d. o. & tremaine , s. t. 1984 , , 286 , 27 rood , h. j. , page , t. l , kintner , e. c. & king , i. r. 1972 , , 175 , 627 schade , d. , carlberg , r. g. , yee , h. k. c. , lpez - cruz , o. & ellingson , e. 1996a , , 464 , l63 schade , d. , carlberg , r. g. , yee , h. k. c. , lpez - cruz , o. & ellingson , e. 1996b , , 465 , l103 schwarzschild , m. 1954 , , 59 , 273 shaya , e. j. , peebles , p. j. e. , & tully , r. b. 1995 , , 454 , 15 smith , s. 1936 , , 83 , 29 strauss , m. a. & willick , j. a. 1995 , , 261 , 271 the , l. s. & white , s. d. m. 1984 , , 92 , 1248 tremaine , s. , richstone , d. o. , byun , y - i . , dressler , a. , faber , s. m. , grillmair , c. , kormendy , j. , & lauer , t. r. 1994 , , 107 , 634 tsai , j. c. & buote , d. a. 1996 , , 458 , 27 van albada , g. b. 1960 , , 15 , 165 walker , t. p. , steigman , g. , kang , h. , schramm , d. m. & olive , k. a. 1991 , , 376 , 51 west , m. j. & richstone , d. o. 1988 , , 335 , 532 white , d. a. & fabian , a. c. 1995 , , 273 , 72 white , s. d. m. 1976 , , 177 , 717 white , s. d. m. , navarro , j. f. , evrard , a. e. , & frenk , c. s. 1993 , , 366 , 429 yee , h. k. c. , ellingson , e. & carlberg , r. g. 1996 , , 102 , 269 yee , h. k. c. , sawicki , m. j. , ellingson , e. , & carlberg , r. g. 1995 , in _ fresh views on elliptical galaxies _ asp conference series , 86 , 301 zembrowski , p. & carlberg , r. g. 1996 , in preparation zwicky , f. 1933 , helv .
acta 6 , 110 zwicky , f. 1937 , , 86 , 217
|
the systematic errors in the virial mass - to - light ratio , @xmath0 , of galaxy clusters as an estimator of the field @xmath1 value are assessed .
we overlay 14 clusters in redshift space to create an ensemble cluster which averages over substructure and asymmetries .
the combined sample , including background , contains about 1150 galaxies , extending to a projected radius of about twice 200 .
the radius 200 , defined as where the mean interior density is 200 times the critical density , is expected to contain the bulk of the virialized cluster mass .
the dynamically derived @xmath2 of the ensemble is @xmath3 .
the @xmath0 overestimate is attributed to not taking into account the surface pressure term in the virial equation . under the assumption that the velocity anisotropy parameter is in the range @xmath4
, the galaxy distribution accurately traces the mass profile beyond about the central @xmath5 .
there are no color or luminosity gradients in the galaxy population beyond @xmath6 , but there is @xmath7 mag fading in the @xmath8 band luminosities between the field and cluster galaxies .
we correct the cluster virial mass - to - light ratio , @xmath9 ( calculated assuming @xmath10 ) , for the biases in @xmath11 and mean luminosity to estimate the field @xmath12 . with our self - consistently derived field luminosity density , @xmath13 ( at @xmath14 ) , the corrected @xmath1 indicates @xmath15 ( formal @xmath16 random error and estimated potential systematic errors ) for those components of the mass field in rich clusters . 1 200 200
|
the main objective of our observations was to obtain physical parameters for a selection of metal - rich globular clusters in m31 .
these parameters are usually straight - forward to derive from color - magnitude diagrams ( cmds ) , however , we have found the effects of crowding to be particularly severe .
we therefore present our analysis in two parts .
this paper describes in detail the effects of crowding , how we quantify them , the techniques used to correct for them , and their implications for future space - based observations such as with ngst . a second paper ( * ? ? ?
* hereafter paper ii ) presents the science obtained from our observations using these techniques .
the general effects of crowding are well known , although not always accounted for .
these effects include reduced photometric accuracy which broadens the cmd and smears the luminosity function ( lf ) , reduced positional accuracy @xcite , artificial brightening and shifts in the measured colors @xcite , and in severe cases , the creation of objects brighter than anything in the parent population through random groupings of many stars @xcite .
if these effects are not taken into account , they can result in a mistaken spread in metallicity assumed from the rgb width , an incorrect mass function extrapolated from the lf , or a false age estimation from the agb peak luminosity .
this paper is organized as follows .
section [ sec : observations ] presents the details of our observations , and section [ sec : data_reduction ] explains the reduction procedures .
section [ sec : artificial_star_tests ] describes the different simulations in detail .
section [ sec : results ] presents the results which are applied to the cluster data in paper ii , and section [ sec : implications ] shows how these results can be applied to future space - based observations , such as the ngst .
we summarize our main conclusions in section [ sec : conclusions ] .
we have obtained hst nicmos images of five of m31 s metal rich globular clusters and their surrounding fields ( cycle 7 ; program i d 7826 ) .
the clusters are g1 , g170 , g174 , g177 & g280 ; however most of the discussion in this paper will focus on g280 which has the densest core , one of the sparsest surrounding fields , and is dispersed enough to allow accurate cluster star measurements in its outer regions .
our observations were taken with the nicmos camera 2 ( nic2 ) which has a plate scale of @xmath0 00757 pixel@xmath1 and a field of view of 194 on a side ( 376 arcsec@xmath2 ) .
the nicmos focus was set at the compromise position 1 - 2 , which optimizes the focus for simultaneous observations with cameras 1 and 2 .
all of our observations used the multiaccum mode @xcite because of its optimization of the detector s dynamic range and cosmic ray rejection .
each of our targets was observed through three filters : f110w ( 0.81.4 ) , f160w ( 1.41.8 ) , and f222 m ( 2.152.30 ) .
these filters are close to the standard ground - based @xmath3 , @xmath4 , & @xmath5 filters . each globular cluster
was observed over three orbits of hst , with @xmath6 minutes of observations per orbit .
this yielded total integration times of 1920s in f110w , 3328s in f160w , and 2304s in f222 m ( see table [ tab : filters ] ) .
we implemented a spiral dither pattern with 4 positions to compensate for imperfections in the infrared array .
the dither steps were 04 for the @xmath3 and @xmath5 band images , and 50 for the @xmath4 band images .
thus the full size of the combined dithered images are @xmath7 in @xmath3 and @xmath5 , and @xmath8 in @xmath4 .
we used the predefined sample sequences step32 with 22 samples in @xmath3 and 25 samples in @xmath5 , and step64 with 21 samples in @xmath4 @xcite .
the @xmath4-band image of g280 is shown in figure [ fig : g280 ] .
this is the combination of 4 dither positions , and is @xmath8 on a side .
our data were reduced with the stsci pipeline supplemented by the iraf nicproto package ( may 1999 ) to eliminate any residual bias ( the `` pedestal '' effect ) . to account for any bias shift which may have occurred in the nicmos reduction ( arbitrary amounts get added or subtracted during pedestal removal ) , or unsubtracted background contribution , we scale our surface brightness by adding a constant to the entire frame to agree with the surface brightness from @xcite . as an example , the procedure for the g280 frame was as follows . taking the observed position of g280 as @xmath9 from the center of m31 , and 34 from the major axis , we interpolate kent s data to find an equivalent major axis distance of @xmath10 from m31 s center . at this distance
the total @xmath11-band surface brightness is @xmath12 magnitudes arcsecond@xmath13 .
assuming an @xmath14 color of 2.9 from @xcite for the disk of a galaxy similar to m31 , we estimate the @xmath5-band surface brightness to be 18.3 magnitudes arcsecond@xmath13 .
we then added a constant 0.045 to the count - rate in each pixel of our frame to bring the average surface brightness of the frame up to this level ( measured @xmath15 around the cluster ) .
object detection was performed with daofind on a combined image made up of all the dithers of all the bands ( 12 images in total ) .
psfs were constructed with daophot for each of the four dithers , then averaged together to create a single psf for each band of each target ( the average fwhm of each band is listed in table [ tab : filters ] ) .
instrumental magnitudes were measured using the allframe psf fitting software package @xcite , which simultaneously fits psfs to all stars on all dithers .
daogrow @xcite was used to determine the best magnitude in a @xmath16 radius aperture , which we then converted to the cit / ctio system using the transformation equations of @xcite .
cccc f110w & 1920 & 1.65 & 0.13 + f160w & 3328 & 1.95 & 0.15 + f222 m & 2304 & 2.45 & 0.19 + [ tab : filters ]
the measurements of stars in the m31 clusters and their surrounding fields will be affected by crowding due to the high stellar density . we have performed three types of tests to evaluate the effects of crowding on our photometry and to quantify at what surface brightness levels we can achieve accurate measurements .
we compare these techniques with one another ( [ sec : comparison ] ) , and with predictions from first principles ( [ sec : simulations_vs_theory ] ) . our first attempt to determine the effects of crowding , as well as to quantify our photometric completeness , used traditional artificial star tests .
we added stars of a fixed magnitude to each individual dither using addstar @xcite , re - processed the frames , and recorded the recovered magnitudes .
since we are interested in not only the completeness , but also the effects of blending , we did not incorporate a delta magnitude criterion for the recovered stars .
we repeated this procedure for nine different input magnitudes @xmath17 .
each trial added only 81 stars per frame to minimize additional crowding .
these 81 stars were arranged in a grid with spacing of @xmath18 pixels to avoid self - crowding .
the grid of input stars was then shifted by 3 pixels and the process repeated so that the entire cluster was covered in 36 trials for each of the 9 input magnitudes ( in the end an artificial star of each magnitude had been added to every 9th pixel in the region of the cluster ) .
we thus input 2916 artificial stars per magnitude .
not surprisingly , figure [ fig : completeness ] demonstrates that the completeness is a function of both input magnitude and distance from the cluster center .
the top panel shows the fractional completeness for the entire g280 frame as a function of input magnitude . here
each bin represents a total of 2916 artificial stars added 81 stars at a time .
the error bars represent the dispersion in the number of recovered stars over the 36 trials .
the bottom panel shows the completeness of each magnitude trial as a function of position from the center of the cluster . here
the faintest input magnitude plotted is @xmath19 , which is only @xmath20% complete in the outer regions , and drops to a value consistent with zero at @xmath21 .
input magnitudes from @xmath22 to @xmath23 inclusive , show essentially 100% completeness for radii @xmath24 .
input magnitudes @xmath25 and @xmath23 remain 100% complete down to @xmath26 radius figure [ fig : artstarfits ] illustrates an important aspect of the crowding problem not shown by figure [ fig : completeness ] ; the difference between input and recovered magnitudes . as will be seen , for our purposes , the traditional completeness plots ( fig [ fig : completeness ] ) are largely irrelevant .
we are mainly concerned with accurately measuring the upper red giant branch ( rgb ) , and thus blending at the bright end , and not completeness at the faint end , is our primary worry .
figure [ fig : artstarfits ] shows the effects of blending as a function of radial distance from the cluster .
the dots are the measured objects in the g280 frame .
the lines are the locus of the recovered artificial star magnitudes . at large radii @xmath27
there is good correspondence between the input and the recovered ( plotted ) magnitudes ( thus the actual value of each set of input magnitudes can be obtained from the level of each line at @xmath28 ) .
however at small radii , what is measured can be several magnitudes above what was input .
regardless of the star s input brightness , if a star is recovered near the center of the cluster , it is recovered with the luminosity of the object on which it fell ( if it does not fall near a bright star , then it is most likely not recovered ) .
note that the brightest objects measured on the real frame are all within the central few arcseconds , and that the upward curving loci of the artificial stars closely mimic the distribution of those bright stars measured on the real frame .
this is indicative of the blended nature of these objects , but does not reveal their true luminosities , as _ all _ the loci of the artificial stars pass through the brightest measured objects at the center of the cluster . from the photometric results in figure [ fig : artstarfits ] , we see that we measure predominantly very bright objects in crowded regions .
the artificial star tests showed that the measurement of faint stars injected into these crowded regions will be significantly altered by the severe crowding , and these faint stars will usually be absorbed into the nearest bright object .
to further investigate the nature of these bright objects , we decided that more tests were necessary .
however , heeding the warning of @xcite that typical artificial star experiments can not reproduce the true effects of severe image crowding , and following the lead of @xcite , we initiated our own set of `` non standard '' artificial star experiments .
these experiments involved the simulation of the entire cluster and surrounding field using a reasonable estimate of the expected cluster star properties . starting with a blank frame with the appropriate noise characteristics , we randomly added cluster and field stars , using the same psfs determined from the g280 cluster , with any negative values set to zero .
both the cluster and field stars follow the luminosity function and colors observed in the galactic bulge , as we expect the cluster stars to have similar properties , and the bulge lf is complete to faint levels in the ir .
the lf is a power law with a slope of 0.278 , extending from @xmath29 ( @xmath30 at baade s window ) @xcite .
later we will examine the effect of changes in the assumed input lf .
colors are derived from a combination of observations of the rgb slope for @xmath31 @xcite , and from stellar models for @xmath32 ( cassisi ) .
cluster stars were added according to the radial surface brightness profile measured in the real cluster until the integrated magnitude of the cluster matched that of the cluster being modeled .
note that the measurement surface brightness is independent of crowding , and as long as there are not saturated pixels , which there were not , the measured sb is accurate .
field stars were added randomly to approximately match the observed field star density .
the artificial frames were then processed and measured in exactly the same manner as the real data . in order to best understand the effects of crowding , we focus on the g280 observations which have the largest contrast in surface brightness between field and cluster .
g280 has one of the densest cores , is one of the most extended clusters in our sample , and at the same time , has one of the sparsest fields .
a280 , the artificial analog of the g280 cluster , is shown in figure [ fig : a280 ] .
this frame is composed of 450,000 cluster stars and 80,000 field stars .
the easiest way to differentiate between the g280 ( fig .
[ fig : g280 ] ) and the a280 ( fig .
[ fig : a280 ] ) frames is to look for the presence ( or absence ) of bad pixels , of which there are none on the artificial frame .
figure [ fig : a280lf ] shows the input ( dashed ) and measured ( solid ) luminosity functions for the a280 frame .
the top panel is for the whole frame .
even though the input lf cuts off at @xmath33 , the measured lf extends to @xmath34 . assuming that most of the recovered stars within @xmath35 belong to the cluster , we break the lf up into the cluster and field contributions .
the middle panel shows the lf for the field around g280 , only counting objects located greater than @xmath36 from the cluster center . here
the measured lf shows very good correspondence with the input lf , having identical bright end cutoffs .
the faint end decline at @xmath37 is due to `` classical '' incompleteness .
the bottom panel includes only objects closer then @xmath36 from the cluster center ; we assume that these are mostly cluster stars .
this is where the major discrepancy between the input and recovered lf occurs .
all of the stars measured brighter than the input lf cutoff , i.e. brighter than any `` real '' star , are found in this small region of the frame .
we emphasize that the same lf was used for the entire frame .
the a280 cmds are shown in figure [ fig : a280cmd ] where we have again divided the field spatially .
objects closer than @xmath36 from the cluster center are plotted in the left panel , while objects farther than @xmath36 are in the right panel .
points and circles indicate measured objects ; the locus of the input stars is indicated by the line . in the right panel
there is a very good correspondence between the input and recovered magnitudes .
there is very little crowding , and as a result very little scatter of the measured data around the input data .
the scatter at the faint end gives an idea of the scale of the measurement errors . in the left panel
, however , there is a large plume of stars measured brighter and bluer than anything input into the cluster .
this blue color is an immediate indicator of the blended origin of these objects , where what is measured is not a single bright star , but the superposition of many fainter , bluer stars . using criteria developed in
[ sec : results ] , we plot objects inside the threshold - blending radius , where the blending is just beginning to affect the faintest stars , at surface brightness of @xmath38 magnitudes arcsecond@xmath13 , with open circles .
objects measured inside the critical - blending radius , where the surface brightness is @xmath39 and blending has severely distorted nearly all measurements , with half - size dots .
note that nearly all recovered stars lie systematically to the blue of the input line for @xmath40 .
this suggests that blending is affecting all stars to some degree , which is not surprising since @xmath41 is also the dividing line inside which there were added more than one star per pixel .
the g280 @xmath42-@xmath43 cmds are shown in figure [ fig : g280cmd ] for comparison .
the left panel shows all the data inside a radius of @xmath36 , the distance chosen to define the cluster .
objects inside the threshold - blending limit @xmath44 are plotted with open circles , and objects inside the critical - blending limit @xmath45 ) are plotted with half - size dots .
these radii are determined in
[ sec : results ] , and listed in table [ tab : blend_radii ] .
the right panel shows all objects outside the @xmath36 cluster radius .
these stars are expected to be non - cluster , or `` field '' stars .
the color spread in the g280 field cmd is real .
this field is @xmath46 disk and @xmath47 bulge , with an estimated spread in [ fe / h ] of @xmath48 ( paper ii ) . the artificial a280 frame , in contrast ,
was created using a single metallicity , and thus shows a much smaller color spread .
the field subtracted cluster cmd of g280 also looks remarkably similar to the a280 cluster cmd ( paper ii ) .
table [ tab : a280blends ] summarizes the true blending of stars over a range in measured luminosities in the a280 cluster .
we have randomly chosen stars with approximately integer measured magnitudes , and assessed their environment and number of neighbors which are contributing to their measured flux .
the first three columns list the measured absolute @xmath5-band magnitude , the distance of the star from the center of the cluster , and the average surface brightness at that distance respectively .
the fourth column lists the number of stars which were input within a 2 pixel ( @xmath49 ; about the size of one resolution element ) radius of the measured star s position , and the last column the number of stars brighter than @xmath50 within 2 pixels .
even though there are _ many _ faint stars within 2 pixels of _ most _ of the measured stars , the number of such stars which are relatively bright is much less .
ccccc -8.94 & 0.38 & 12.3 & 12777 & 88 + -8.01 & 0.73 & 13.4 & 4233 & 36 + -7.00 & 1.10 & 14.4 & 1661 & 11 + -5.99 & 1.53 & 15.5 & 644 & 6 + -5.00 & 1.71 & 15.9 & 451 & 5 + -4.01 & 2.61 & 17.7 & 87 & 1 + -3.01 & 9.48 & 20.2 & 10 & 1 + [ tab : a280blends ] we have run simulations varying the shape of the lf input into the artificial cluster . the results are summarized in table [ tab : variable_lfs ] .
the first two columns list the limits of the power - law input lf , and the third column is the slope @xmath51 ( @xmath52 ) .
the fourth column gives the number of stars input into the cluster ; this number was scaled in order to keep approximately the same integrated cluster magnitude .
the last two columns list the brightest star measured , and the average @xmath43 color of the five brightest stars in each cluster .
note that the `` normal '' a280 lf is taken from baade s window with a slope of 0.278 extending from @xmath53 , and has 450,000 stars .
the top half of table [ tab : variable_lfs ] describes the simulations where we varied the bright end extent of the input lf by @xmath54 magnitudes .
all the simulations achieve approximately the same core surface brightness , and , as a result , generate blended objects which all reach nearly the same brightness .
the most noticeable difference between these simulations is the recovered colors .
the simulation with the faintest input lf cutoff ( and hence the most stars ) , returns stars with bluer colors , by as much as @xmath55 at the bright end .
the bottom half of table [ tab : variable_lfs ] is for the simulations where we varied the input lf slope by @xmath56 .
as before , all simulations achieve approximately the same core surface brightness , and blends up to @xmath57 .
the @xmath58 trial has a noticeably wider rgb due to the many blends with fainter stars .
however , the brightest blends do not show the same gradual progression towards bluer colors with increased number of input stars as was observed when varying the input lf cutoff . can these simulations recover the true lf from a blended frame ?
in the fields surrounding the clusters , the simulations confirm that what we measure is indeed the true lf .
however , in the cores of the clusters where the stars are severely blended , there is a degeneracy between the combination of lf extent , slope , and number of stars ( surface brightness ) . in these dense regions
it is nearly impossible to distinguish between a cluster with 250,000 stars going as bright at @xmath59 , and another cluster with stars as bright as @xmath60 but three times as many stars , at least using the current observations and techniques .
cccccc 5.8 & -9.2 & 0.278 & 254000 & -9.85 & 1.11 + 5.8 & -8.2 & 0.278 & 338000 & -9.92 & 1.03 + 5.8 & -7.2 & 0.278 & 450000 & -9.66 & 0.94 + 5.8 & -6.2 & 0.278 & 601000 & -9.93 & 0.87 + 5.8 & -5.2 & 0.278 & 805000 & -9.41 & 0.72 + 5.8 & -7.2 & 0.238 & 207000 & -9.74 & 1.12 + 5.8 & -7.2 & 0.258 & 306000 & -9.77 & 0.93 + 5.8 & -7.2 & 0.278 & 450000 & -9.66 & 0.94 + 5.8 & -7.2 & 0.298 & 653000 & -9.61 & 0.93 + 5.8 & -7.2 & 0.318 & 933000 & -9.69 & 0.93 [ tab : variable_lfs ] we now compare the luminosity functions from the traditional completeness tests , where a handful of artificial stars were added to an observed frame , to those from the artificial cluster tests .
we start with the traditional completeness test lf shown in the top panel of figure [ fig : artstarlf ] .
the dashed line illustrates the input lf composed of 2916 input stars per magnitude , achieved 81 stars at a time over 36 trials .
the solid line is the recovered lf . to mimic a power - law input lf
, we multiply the input lf by the same power - law slope as was used in the artificial cluster tests ( @xmath61 ) . the number of stars recovered ( at any magnitude ) is then multiplied by the same constant that was applied to the responsible input magnitude . the recovered stars from each input magnitude
are then summed to get the total recovered lf .
the resulting input ( dashed ) and recovered ( solid ) lfs are shown in the middle panel of figure [ fig : artstarlf ] .
however , from our initial completeness tests , we input stars several magnitudes brighter than the true extent of the lf .
( this is because with just the traditional completeness tests , the true brightnesses of the measured stars are unknown , because one is just placing artificial stars on the preexisting clumps of stars . )
thus we now truncate the input lf at @xmath62 to ( nearly ) match the artificial cluster input lf .
the results are shown in the bottom panel of fig .
[ fig : artstarlf ] . the traditional completeness test results ( bottom fig .
[ fig : artstarlf ] ) are approximately the same as the artificial cluster results ( top fig .
[ fig : a280lf ] ) .
the input lfs ( dashed lines ) start at @xmath63 and follow a power - law distribution upward toward fainter magnitudes .
the observed lf ( solid line ) extends three magnitudes brighter than the input lf , and drop off at @xmath64 due to incompleteness . at the faint end of the photometry ,
both tests are equally useful to characterize the photometric completeness .
the artificial clusters simulations are , however , a more controlled experiment that offer advantages to the interpretation of the nature of the brightest measured stars .
the traditional completeness tests show that some stars are brightened , but when the crowding gets severe , the brightening seems to come from falling on or near a preexisting bright clump , and measuring the brightness of the clump rather than the individual star .
these tests , however , do not tell you whether the bright clumps are real . the artificial cluster simulations , on the other hand ,
tell you the number and brightness of the stars which make up the clumps , and in our case , that these bright objects which are observed can be created with a normal lf whose brightest star is several magnitudes fainter than the brightest object observed .
we have created uniformly populated mini - fields of @xmath65 pixels to investigate any possible effects of the strong surface brightness gradient present in the observed globular clusters .
this is especially important since we use surface brightness as the criterion for quantifying blending , and determining the believability of our photometry .
each mini - field is constructed in the same manner as the artificial clusters , using a power - law input luminosity function , extending across @xmath66 . between @xmath67 and @xmath68 stars
are randomly added to each field , where the field with the most stars nearly reaches the surface brightness level observed in the cores of the m31 clusters .
the input ( dashed lines ) and recovered ( solid lines ) luminosity functions of these fields are shown in figure [ fig : artfieldlfs ] . the surface brightness of each field is listed in the upper right corner of each plot .
there is good agreement between the bright end extent of the input and measured lfs up to a surface brightness of @xmath69 .
however as more and more stars are added to the field , blending becomes significant , the lf shifts toward brighter and brighter magnitudes , and then more stars are `` recovered '' with @xmath42 brighter than the brightest input star .
notice the change of the slope of the measured lf .
when blending is insignificant , the measured lf is very close the the input lf , i.e. increasing toward fainter magnitudes with the correct slope .
however when blending becomes severe , more stars clump together , and fewer faint stars are recovered .
the relative number of bright stars compared to faint stars changes dramatically , and the slope of recovered lf now has an opposite sign compared to the input lf .
our tests show that the surface brightness gradient appears inconsequential , insofar as the effects of blending on photometry are concerned .
an advantage of the mini - fields over the artificial clusters is that they give a larger sample of stars at each surface brightness interval .
this makes it easier to quantify at what surface brightness levels the photometry is either accurate , tainted , or corrupt .
we find that many of the bright objects observed in crowded regions arise from the clumping of many fainter stars .
although the traditional completeness tests do show that faint stars are artificially brightened , they do not give the whole picture .
our artificial frames , which closely mimic the real hst - nicmos data , have shown that random clumps of stars can create objects several magnitudes brighter than the brightest real stars .
figure [ fig : artfieldsb ] , a plot of the recovered absolute @xmath5-band magnitude as a function of the field surface brightness , shows the results of this blending .
here each curve represents the median of stars input with a range of @xmath70 magnitudes .
for example the bottom - most curve gives the median recovered @xmath42 of all stars input with magnitudes between @xmath71 and @xmath72 in each @xmath65 pixel mini - field ( 76 @xmath73 76 ) .
the number of stars in each 1 magnitude range is a function of the total number of stars input ( i.e. the surface brightness ) , and the range being considered , since we use the same power - law input luminosity function .
thus the statistical noise is quite high at the faint surface brightness end , especially for the brighter input curves .
note that there were no stars at @xmath74 or @xmath75 at @xmath76 , or for @xmath77 at @xmath78 .
figure [ fig : artfieldsb_deltamag ] shows a plot of the average deviation between the input and recovered magnitudes for stars of different input brightnesses , as a function of the average surface brightness of the artificial field .
this plot illustrates how increasing surface brightness has an increasingly strong effect on stellar photometry . at surface brightnesses fainter than @xmath79 magnitudes
arcsecond@xmath13 , all stars are measured fairly accurately . however at @xmath80 the fainter stars start to become artificially brightened by crowding . with increasing surface brightness ,
the effects of crowding become more and more significant for brighter stars . at @xmath81
the crowding is so severe that nearly all stars are badly blended or misidentified with random clumps of blended stars , and even the brightest input stars are mismeasured by several magnitudes . comparing the fits to the recovered artificial star magnitudes from the traditional completeness tests ( figure [ fig : completeness ] ) , and the recovered magnitudes from the artificial mini - fields ( figure [ fig : artfieldsb_deltamag ] ) ,
the same effect is apparent . at some critical surface brightness , or radius in the case of our globular clusters ,
blending effects begin to dominate . as the surface brightness increases ,
blending becomes more and more important , until even the brightest stars become lost in the uneven background ( for more on the effects of crowding on positional errors see @xcite ) .
what traditional completeness tests ( adding stars to an existing frame ) , do not show is the true upper limit to the lf .
for example , look at the plot of recovered magnitudes from the artificial mini - fields ( figure [ fig : artfieldsb ] ) .
suppose an object is measured at @xmath82 in a region where the surface brightness is @xmath83 .
this object could be a star of @xmath82 , or it could just as easily be a blend of many stars several magnitudes fainter . using figures [ fig : artfieldsb ] and [ fig : artfieldsb_deltamag ] as guides
, we chose @xmath84 as the threshold point where our photometry starts to become noticeably affected by blending , and @xmath85 as the critical blending radius where our photometry is dominated by blends . beyond this level ,
no measurements are reliable .
figure [ fig : surface_brightness ] shows azimuthally averaged surface brightness profiles of each cluster ( surface brightness values have been normalized from @xcite photometry as described in [ sec : data_reduction ] ) .
dotted lines indicate the threshold- and critical - blending surface brightness levels .
we use this plot to determine the threshold - blending radius ( @xmath86 ) , and the critical - blending radius ( @xmath87 ) for each cluster .
these radii are listed in table [ tab : blend_radii ] .
any objects measured inside the threshold - blending radius , especially faint objects , are potentially affected by blending , and should be considered suspect . (
note that this radius ( @xmath86 ) was chosen so that stars input at @xmath88 could be recovered accurately most of the time . however , for stars fainter than this , there is no way to tell whether stars measured at @xmath89 are blends or not . )
objects measured inside the critical - blending radius are undoubtedly blends , and although we plot them for completeness , they should be disregarded .
cccc g001 & 11.2 & 1.2 & 2.9 + g170 & 12.9 & 0.5 & 1.4 + g174 & 11.7 & 0.5 & + g177 & 11.8 & 0.6 & + g280 & 10.7 & 1.0 & 2.2 [ tab : blend_radii ] in summary , traditional completeness tests are good for analyzing one s detection efficiency and photometric errors . however , in the case of blended images , these tests can not reveal what the blends are composed of . only by creating completely artificial frames , and matching the observed and modeled surface brightnesses
, can one get an idea of which stellar populations are consistent with the observations .
even then there are degeneracies between the extent of the lf , the slope of the lf , and the true number of stars . since in paper
ii we will use the slope of the rgb to estimate metallicities , and the width of the rgb to place limits on any spread in metallicity , we have investigated the effects of crowding on these quantities .
this specific case should serve as a good example for other , similar analyses .
this analysis uses the mini - fields discussed in the previous section , which include between @xmath67 and @xmath68 stars , achieving surface brightnesses in the range @xmath90 .
the input @xmath42-@xmath43 rgb slope is @xmath91 , which implies [ fe / h ] @xmath92 according to the rgb slope
metallicity relation of @xcite .
we perform an iterative linear least squares fit to the measured rgb , rejecting measurements farther than @xmath93 from the fit .
even though the metallicity relation was derived for the rgb in the range @xmath94 , here we fit to all the data since blending shifts the rgb several magnitudes in luminosity .
figure [ fig : gbslope ] shows the rgb slope measured in each simulated mini - field as a function of the average field surface brightness .
the dotted horizontal line at a slope of @xmath91 shows the input rgb slope , i.e. the slope which should be recovered in every field if crowding were unimportant .
indeed at the low surface brightness end , the recovered rgb slope approaches the input rgb slope .
as the surface brightness of the field increases @xmath95 , more faint blue stars clump together pushing the lower end of the rgb bluer and to higher luminosities , thus tilting the rgb which implies higher metallicities . at even higher surface brightnesses
@xmath96 , the blending now also significantly affects even the brightest stars .
all the faint stars are lost in the background , and the only objects measured are just the brightest stars clumped together with many fainter bluer stars . the dashed line is a linear fit to the data with @xmath97 , and is the relation we use to estimate corrections to our measured rgb slopes .
we emphasize that this correction is merely an estimate of the effects of crowding .
we do not plot the correction as a function of delta slope or delta [ fe / h ] , in order to discourage blind use , as the effects of blending will be different depending on the specific input luminosity function and colors . on the right side of figure
[ fig : gbslope ] is the metallicity scale implied from the relation of @xcite .
this shows the strong dependence of the metallicity on the slope , and thus the severe effect that even minimal blending can have on the metallicity derived from the rgb slope .
we now analyze the effects of blending on the width of the measured rgb .
all of our simulations start out with an input rgb of zero width . to measure the width of the rgb
, the linear fit to the rgb is subtracted off to leave a vertical rgb centered at zero @xmath43 color , and the distribution is binned into 0.05 magnitude color bins .
we then fit a gaussian to the binned distribution .
figure [ fig : gbwidth ] shows the width of the gaussian fit to the measured giant branch of each field as a function of the field s average surface brightness .
this plot shows that the measured width of the rgb is not significantly affected by crowding until the surface brightnesses exceeds @xmath98 . at that point
the width grows approximately linearly with increasing surface brightness . a great deal on what to expect from crowding can be derived from first principles @xcite . in this section the main points in this theoretical approach
are summarized and then compared to the results of our simulations . for a given stellar population of age @xmath99 and total bolometric luminosity @xmath100 the number @xmath101 of stars in the generic @xmath102 ( post main sequence ) evolutionary stage of duration @xmath103 is given by @xmath104 where @xmath105 is the specific evolutionary flux , i.e. the number of stars leaving or entering any evolutionary stage per year and per unit ( solar ) luminosity of the population . in practice , @xmath105 is only weakly dependent on age , varying from @xmath106 to @xmath107 ( stars l@xmath108@xmath1yr@xmath1 , see fig . 1 in * ? ? ?
equation ( 1 ) can be applied to one individual resolution element , in which case the sampled luminosity can be written as @xmath109 where @xmath110 is the linear size of the resolution element ( in arcsec ) , @xmath111 is the surface brightness of the stellar system in units of l@xmath108@xmath112 , @xmath113 is the surface brightness in units of l@xmath108@xmath114 , and @xmath115 is the distance . for @xmath116 , @xmath117 is close to the probability that a generic resolution element contains one star in the @xmath102 phase .
therefore , @xmath118 is the probability that a resolution element contains a blend of two such stars that a photometric package will mistake for a single star about twice as bright .
more generally , the probability that a resolution element will contain a blend of a star in phase @xmath102 and a star in phase @xmath119 is given by : @xmath120 the bottom line is that , for given physical crowding as measured by the surface brightness ( in l@xmath108/pc@xmath2 units ) , the number of two - star blends is proportional to the fourth power of the linear size of the resolution element , to the fourth power of the distance from the observer to the stellar population , and to the square power of the surface brightness @xcite .
for example , suppose we have observed a globular cluster at a distance of 7 kpc with a resolution of @xmath121 , and would like to obtain the same frequency of two - star blends for an identical cluster in m31 , i.e. at a distance of 700 kpc .
equation ( 3 ) tells that we would need a resolution 100 times better , i.e. @xmath122 .
for example , equations ( 2 ) and ( 3 ) can be used to estimate the average number of rgb stars ( @xmath123 ) , and rgb stars within one magnitude of the rgb tip ( @xmath124 ) , per resolution element .
these can then be used to determine the number of resolution elements in a frame that contain a blend of two rgb stars ( @xmath125 ) , two rgbt stars ( @xmath126 ) , three rgbt stars ( @xmath127 ) , etc .
note that the number of triplets scales as the cube of @xmath124 , that of quadruplets as its fourth power , and so on @xcite .
the results of our calculations are listed in table [ tab : theoretical_populations ] . here
we assume a 15gyr population of solar metallicity ( @xmath128 ) , a distance modulus of 24.4 , no @xmath5-band extinction , a 376 arcsec@xmath2 field of view , and we use the @xmath4-band psf fwhm of @xmath49 as the size of our resolution element .
the first column is the @xmath5-band surface brightness ( @xmath129 ) , and the second is the total bolometric luminosity sampled by a single resolution element ( @xmath130 ) .
the third and fourth columns give the average number of stars on the red giant branch ( @xmath131 ) and stars within one magnitude of the red giant branch tip ( @xmath132 ) sampled by a single resolution element .
the fifth , sixth , seventh and eighth columns give the number of 2 rgbt star blends ( @xmath133 ) , 3 rgbt star blends ( @xmath134 ) , two or more rgbt star blends ( @xmath135 ) , and number of 2 rgb star blends ( @xmath136 ) expected on the entire frame .
the last column is the total @xmath137band magnitude sampled by one resolution element .
cc|cc|cccc|c 13 & 5340 & 0.6 & 70 & 7334 & 4308 & 17766 & & -7.02 + 14 & 2126 & 0.2 & 28 & 1162 & 272 & 1517 & & -6.02 + 15 & 846 & 0.09 & 11 & 184 & 17 & 203 & & -5.02 + 16 & 337 & 0.04 & 4 & 29 & 1 & 30 & & -4.02 + 17 & 134 & 0.02 & 2 & 5 & 0.07 & 5 & & -3.02 + 18 & 53 & 0.006 & 0.7 & 0.7 & 0.004 & 0.7 & 10561 & -2.02 + 19 & 21 & 0.002 & 0.3 & 0.1 & 0.0003 & 0.1 & 1674 & -1.02 + 20 & 8 & 0.0009 & 0.1 & 0.02 & 0.0000 & 0.02 & 265 & -0.02 + 21 & 3 & 0.0004 & 0.04 & 0.003 & 0.0000 & 0.003 & 42 & 0.98 + 22 & 1 & 0.0001 & 0.02 & 0.0005 & 0.0000 & 0.0005 & 7 & 1.98 + [ tab : theoretical_populations ] the result is graphically shown in fig .
[ fig : blends ] .
this plot shows the predicted number of blends of 2 rgbt stars ( @xmath138 dashed line ) , and 2 or more rgbt stars ( @xmath139 solid line ) on the entire nic2 frame . over plotted
are the number of stars measured brighter than the maximum input luminosity ( @xmath140 ) in the simulations described in section [ sec : artificial_fields ] .
there is remarkable agreement for @xmath141 .
brighter than this , the simulations are reaching the limit of the maximum number of stars which can be fit on a frame , and hence fall they short of the theoretical prediction .
for fainter surface brightnesses the flattening in the simulations is due to a few objects which are recovered only slightly brighter than @xmath140 , being blends of an rgbt star with a much fainter star .
the general ( qualitative ) criterion for safe stellar photometry in crowded field states that reliable photometry can be obtained only for those stars that are brighter than the average luminosity sampled by each resolution element " @xcite .
the last column in table [ tab : theoretical_populations ] gives the total @xmath137band magnitude sampled in m31 by our resolution element .
a comparison with fig . [ fig : artfieldsb ] shows indeed that the criterion is well confirmed by the detailed simulations .
this fairly obvious criterion has been often violated in the past , resulting in a systematic overestimate in the number ( if any ) of stars brighter than the rgb tip in various stellar systems .
even with the improved resolution of new space - based observatories , measurements of distant objects will still be subject to the limitations of image blending .
since the blending threshold is determined by the number of stars present in a resolution element , we can scale our results according to the telescope psf size and target distance to obtain estimates of the limiting surface brightness for different observations .
we need only assume that we have determined the limiting number of stars per resolution element with our nicmos simulations , and that this limit will be similar for other space - based observatories , i.e. assuming similar noise characteristics , psf structure , and performance as nicmos . as an example , consider measuring cepheid variables in external galaxies .
cepheids are one of the most important tools in determining extragalactic distances ; accurate determinations of their periods and apparent magnitudes are one of the most important techniques for obtaining distances to nearby galaxies . assuming a mean @xmath5-band cepheid luminosity of @xmath82 , our simulations show that nicmos observations of such stars at the distance of m31 , the @xmath5-band surface brightness must be fainter than @xmath142 mag / arcsec@xmath2 to make accurate measurements .
this corresponds to a galactocentric distance of @xmath143 using the @xmath11-band @xcite surface brightness measurements , and assuming @xmath144 for a bulge population @xcite . what about cepheid observations in the virgo cluster ? scaling our m31 @xmath5-band nicmos results to the 15.9 mpc distance of virgo
, we find that the brightest background where nicmos can accurately measure cepheids is @xmath145 .
one example of a virgo spiral is ngc 4548 , which
in fact has been imaged by the hst key project team with wfpc2 @xcite .
the @xmath146-band surface brightness measurements of @xcite , indicate that the key project observations occurred in regions with surface brightness @xmath147 . assuming @xmath148 for a young disk population @xcite , colors typical of swb iii clusters in the lmc in which cepheids are found
, these regions have a @xmath5-band surface brightness of @xmath149 mag / arcsec@xmath2 .
thus the regions imaged with wfpc2 may be too dense for accurate nicmos photometry due to the infrared instrument s larger psf .
now consider future observations such as may be performed with a space - based , 8-meter class telescope like the ngst .
the observations of cepheids in the virgo cluster , which are difficult for nicmos , will now be trivial . again scaling our m31 nicmos observations to the anticipated ngst @xmath4-band psf which will have a fwhm @xmath150
, we find that the limiting @xmath5-band surface brightness for accurately measuring cepheid variables is @xmath151 . in ngc 4548
this corresponds to a distance of only @xmath152 from the nucleus .
more distant targets , such as galaxies in the coma cluster , will of course be more challenging . to scale from nicmos observations in m31 to ngst in coma
, we again assume that the @xmath4-band ngst psf has a fwhm @xmath150 , and that the coma cluster is at a distance of 102mpc .
this yields a limiting surface brightness of @xmath153 for accurately measuring cepheids . extrapolating the sb analysis of @xcite ,
this limiting sb occurs at @xmath154 from the center of m31 ( assuming @xmath155 ) . thus placing m31 at the distance of the coma cluster , we find that the inner @xmath156 , or @xmath157kpc ( assuming @xmath158pc in m31 ) , will be inaccessible to even ngst for accurately measuring cepheids in the infrared .
more detailed calculations should of course be undertaken when planning observations , but these scaling arguments give a rough idea of what to expect in terms of the effects of blending .
ccccc m31 & 0.7 & hst - nicmos ( @xmath5 ) & 0.19 & 15.0 + m31 & 0.7 & ngst & 0.04 & 11.6 + virgo & 15.9 & hst - nicmos ( @xmath4 ) & 0.15 & 21.3 + virgo & 15.9 & ngst & 0.04 & 18.4 + coma & 102 & ngst & 0.04 & 22.4 [ tab : cepheid_blend_limits ]
in order to interpret out hst - nicmos observations of globular clusters in m31 , we have developed techniques for understanding and quantifying the effects of blending on stellar photometry .
traditional completeness tests , specifically the injection and recovery of a handful of artificial stars into a preexisting frame , are fine for estimating the photometric completeness . unfortunately , this type of test does not help in understanding the true brightnesses of stars seen in crowded regions . in very crowded regions , the few added artificial stars are merely added on top of preexisting clumps , with no way of recovering the composition of those clumps .
our solution is to simulate the entire frame .
our cluster simulations , consisting of hundreds of thousands of stars each ( fig .
[ fig : a280 ] ) , revealed that stars as bright as we observe near the cluster centers , can be easily created by the blending of many fainter stars ( fig .
[ fig : a280cmd ] ) .
these simulations can be tailored to nearly any observation , and use any desired input lf to best determine the true properties of the observed stellar population , although where blending is severe , there are degeneracies in the shape of the input lf .
using uniformly populated fields of varying surface brightness , we measured the degradation of photometric accuracy as a function of sb .
we determined that the photometry of faint stars ( @xmath88 ) began to be affected by blending at a surface brightness of @xmath80 magnitudes arcsecond@xmath13 . at a surface brightness of @xmath83 the photometry of even the brightest stars
was affected by blending ( fig .
[ fig : artfieldsb ] ) . with these criteria , we determined threshold- and critical - blending radii for each cluster , which determine the proximity to each cluster where reliable photometry can be achieved ( tab .
[ tab : blend_radii ] ) .
we show the effects of blending on the measured luminosity function . at low surface brightnesses
the measured lf accurately reflects the input lf .
however with increasing sb , the measured lf becomes skewed toward the bright end . at very high
sbs blending dominates the photometry , and the measured lf has a slope opposite in sign to the input lf , actually increasing toward bright stars ( fig .
[ fig : artfieldlfs ] ) .
blending also affects the measured slope and width of the giant branch .
we quantified these effects , and determined a relation between the field sb and the artificial increase in the metallicity as derived from the rgb slope ( fig .
[ fig : gbslope ] ) .
results of this modeling will be applied to the cluster and field observations in paper ii . by scaling our limiting surface brightness imposed by blending ,
we estimate the limitations blending places on future space - based observations . we show that infrared observations of cepheid variables with ngst will be trivial in the virgo cluster , but
may be nearly impossible in the coma cluster .
support for this work was provided by nasa through grant number go-7826 from the space telescope science institute .
rmr acknowledges additional support from nasa contract nag5 - 9431 , awarded in connection with the ngst ad hoc science working group .
we thank peter stetson for supplying and helping us with his photometry package , allframe .
valuable comments from darren depoy , paul martini and marcia rieke were greatly appreciated .
|
astronomers are always anxious to push their observations to the limit basing results on objects at the detection threshold , spectral features barely stronger than the noise , or photometry in very crowded regions . in this paper
we present a careful analysis of photometry in crowded regions , and show how image blending affects the results and interpretation of such data .
although this analysis is specifically for our nicmos observations in m31 , the techniques we develop can be applied to any imaging data taken in crowded fields ; we show how the effects of image blending will even limit ngst .
we have obtained hst - nicmos observations of five of m31 s most metal rich globular clusters .
these data allow photometry of individual stars in the clusters and their surrounding fields . however , to achieve our goals
obtain accurate luminosity functions to compare with their galactic counterparts , determine metallicities from the slope of the giant branch , identify long period variables , and estimate ages from the agb tip luminosity , we must be able to disentangle the true properties of the population from the observational effects associated with measurements made in very crowded fields .
we thus use three different techniques to analyze the effects of crowding on our data , including the insertion of artificial stars ( traditional completeness tests ) and the creation of completely artificial clusters .
these computer simulations have proven invaluable in interpreting our data .
they are used to derive threshold- and critical - blending radii for each cluster , which determine how close to the cluster center reliable photometry can be achieved .
the simulations also allow us to quantify and correct for the effects of blending on the slope and width of the rgb at different surface brightness levels .
we then use these results to estimate the limits blending will place on future space - based observations .
|
low - dimensional or frustrated quantum magnets were revealed to exhibit exotic ground states , magnetic excitations , and quantum phase transitions ( qpts).@xcite for a particular case of the spin - gapped antiferromagnets , the external magnetic field can close the gap in the spectrum , which results in a qpt between a low - field disordered paramagnetic phase and a high - field long - range ordered one .
an intriguing finding is that this ordered phase can be approximately described as a bose - einstein condensation ( bec ) of magnons.@xcite many experimental investigations on the critical properties of the bec - related qpt have been carried out , including the characterizations of the magnon spectrum , the magnetization , the specific heat and the thermal transport , etc.@xcite in a recent study , the heat transport properties of a magnon bec material , nicl@xmath1-@xmath6sc(nh@xmath1)@xmath1 ( dtn ) , were found to display strong anomalies at the qpts and the heat conductivity of the bec state seemed to be much enhanced upon lowering temperatures ( approaching the absolute zero).@xcite this result shows an analogy between the magnon bec and the superfluid of @xmath7he in the aspect of the ability of transporting heat . however , one notable facet is that magnons act as heat carriers only in the direction of the spin chains of this compound , whereas they only scatter phonons in the transverse direction.@xcite so the exchange anisotropy may play the key role in the heat transport of magnetic excitations .
a later experimental work confirmed the main features of transport properties of dtn , but an alternative picture based on the mass renormalization and impurity scattering effects was proposed to explain the thermal transport data.@xcite in any case , the qpts associated with the magnon bec are believed to significantly affect the heat transport properties and the low - energy magnetic excitations provide a substantial contribution to the heat transport . to get the general principals of the heat transport in the magnon bec state , we need to carry out systematic studies on more members of the magnon bec materials .
ba@xmath0mn@xmath1o@xmath2 ( bmo ) is an @xmath8 = 1 quantum spin - dimer system exhibiting the bec of spin degrees of freedom .
it crystallizes in the rhombohedral @xmath9 structure with the pairs of mn@xmath10 3@xmath11 ions ( @xmath8 = 1 ) arranged vertically on the hexagonal layers.@xcite the spins of mn@xmath10 ions of each pair are coupled antiferromagnetically to form spin dimers.@xcite neutron scattering results indicated an intradimer exchange energy @xmath12 = 1.642 mev , an interdimer coupling between mn ions in the same plane @xmath13 = 0.1136 mev , an interdimer coupling between mn ions in the adjacent planes @xmath14 mev , and the next nearest neighbor interdimer interactions between bilayers ( along the @xmath5 axis ) @xmath15 mev.@xcite the strong in - plane spin - dimer interaction results in a spin - singlet ground state , with a spin gap of 1.05 mev to the lowest triplet state and a second larger gap to the quintuplet state .
bmo displays a peculiar phase diagram in magnetic field when the zeeman effect lowers the @xmath16 = 1 triplet states and the @xmath16 = 2 quintuplet states.@xcite corresponding to the closures of two spin gaps with increasing magnetic field , there are two sequential magnetically ordered states and four quantum critical fields : @xmath17 = 8.7 t , @xmath18 = 26.5 t , @xmath19 = 32.5 t , @xmath20 = 47.9 t for @xmath21.@xcite the peculiarity is that the first ordered phase for @xmath22 and the second one for @xmath23 are described as the triplet condensation and the quintuplet condensation , respectively.@xcite furthermore , bmo was found to have a weak single - ion uniaxial anisotropy , characterized by @xmath24 = 0.032 mev.@xcite the phase diagram for @xmath25 is a bit more complicated , with two phases of the triplet condensation.@xcite it was revealed that both the single - ion anisotropy and the geometry frustration play crucial roles in determining the phase diagram.@xcite in this work , we study the thermal conductivity ( @xmath3 ) of bmo single crystal to probe the field - induced qpts and the role of magnetic excitations in the heat transport .
it is found that both @xmath26 and @xmath27 are strongly suppressed in magnetic fields , particularly at the field - induced qpts , which demonstrates that at very low temperatures the magnons mainly act as phonon scatterers even in the long - range ordered state .
the present results indicate that magnon - bec state may not necessarily exhibit large thermal conductivity and the ability of magnetic excitations transporting heat is determined by the magnetic structure and the anisotropic spin exchange .
high - quality single crystals of ba@xmath0mn@xmath1o@xmath2 are grown by a slowly cooling method using naoh as a solvent.@xcite the crystals have a hexagonal - platelet - like shape with large sizes up to 5 @xmath28 5 @xmath28 2.5 mm@xmath29 .
the largest surface is the @xmath4 plane and the thickness is along the @xmath5 axis , determined by using the x - ray diffraction and the laue photograph .
the specific heat is measured by the relaxation method in the temperature range from 0.4 to 20 k using a commercial physical property measurement system ( ppms , quantum design ) .
both the @xmath4-plane and the @xmath5-axis thermal conductivities ( @xmath26 and @xmath27 ) are measured using the conventional steady - state technique . particularly , the data at low temperature and in high magnetic field are taken by using a
one heater , two thermometer " method in a @xmath29he refrigerator and a 14 t magnet.@xcite note that we can only probe the qpts at the first critical field of bmo , due to the limitation of the available field .
mn@xmath1o@xmath2 single crystal in zero field .
the solid line indicates the fit of data using formula ( [ sh ] ) .
inset : the data in a broader temperature range from 0.4 to 20 k.,width=245 ] it is known that , in a magnetic insulator , heat can be carried by phonons and magnetic excitations , and the interaction between them usually induces scattering on heat carriers , which is a negative effect on the heat transport . to analysis the heat transport properties quantitatively ,
it is necessary to know the phononic specific heat from the experiments .
figure 1 shows the low-@xmath30 specific heat of the ba@xmath0mn@xmath1o@xmath2 single crystal . at @xmath31 6 k
, there appears a broad shoulder - like feature , which is apparently of magnetic origin .
below 5 k , the specific heat decreases quickly with temperature and it does not show a @xmath32 dependence of the phonon specific heat . the same result has been reported in an earlier work.@xcite the low-@xmath30 specific heat was found to be able to described by a formula for the gapped magnetic excitations,@xcite @xmath33 where @xmath34 is the number of excited states per spin dimer , @xmath35 is the gas constant , and @xmath36 is the energy gap of magnetic excitations .
the low-@xmath30 data can be fitted rather well to eq .
( [ sh ] ) with parameters @xmath34 = 1.49 and @xmath36 = 14.2 k , except for the very low-@xmath30 data ( @xmath37 1 k).@xcite note that the size of gap is quite consistent with those from other measurements.@xcite however , it is difficult to separate the phononic specific heat because of the significant contribution from the magnetic excitations . actually , adding a @xmath32-term to eq .
( [ sh ] ) can not achieve perfect fitting to the data , particulary for the subkelvin data . and @xmath27 of ba@xmath0mn@xmath1o@xmath2 single crystals in zero and @xmath38 t fields .
the dashed lines indicate the approximate @xmath39 and @xmath40 dependencies of the thermal conductivity at very low temperatures and at high temperatures , respectively.,width=321 ] figure 2 shows the temperature dependencies of @xmath26 and @xmath27 of bmo single crystals in zero and 14 t fields , which are applied along either the @xmath4 plane or the @xmath5 axis .
apparently , the zero - field thermal conductivities show rather weak anisotropy at high temperatures .
each @xmath41 curve exhibits a phonon peak at 18 k with the peak values nearly the same for @xmath26 and @xmath27 .
the @xmath40 dependence of @xmath41 at high temperatures is the characteristic of phonon heat transport dominated by the phonon - phonon umklapp scattering.@xcite at sub - kelvin temperatures , both @xmath42 and @xmath43 show an approximate @xmath39 dependence , which indicates that the boundary scattering limit of phonon thermal conductivity is approached or that the microscopic scattering on phonons is negligible.@xcite besides these usual behaviors belonging to the phonon heat transport , there is a shoulderlike " feature in the zero - field @xmath41 at 6.5 k , which is likely due to some kind of resonant scattering on phonons.@xcite applying 14 t magnetic field leads to a strong suppression of low-@xmath30 thermal conductivity and the disappearance of the shoulderlike " feature .
therefore , the phonon resonant scattering in zero field must be of magnetic origin .
similar result has been found in another spin - gapped material , dtn.@xcite on the other hand , there is no evidence for the magnon heat transport in bmo , which is reasonable at very low temperatures considering the negligible magnon excitations due to the finite spin gap . in high magnetic field ,
the strong suppression of phonon heat transport in a broad temperature range is apparently due to the enhancement of magnetic scattering , which can also be seen from the following magnetic - field dependencies of @xmath3 .
another peculiar feature of 14 t @xmath41 data is the small jumps of conductivity at low temperatures , which are related to the field - induced antiferromagnetic ( af ) transitions .
note that even with these increases of @xmath3 at the phase transitions , the lower-@xmath30 thermal conductivities are still much smaller than the zero - field values , indicating that the phonons are still strongly scattered by magnetic excitations in the field - induced af state .
mn@xmath1o@xmath2 single crystals at low temperatures.,width=321 ] figure 3 shows the detailed magnetic field dependencies of @xmath44 and @xmath26 for both @xmath45 and @xmath46 .
it is clear that they are nearly isotropic on the field direction .
the overall behavior of @xmath47 is that the magnetic field strongly suppresses thermal conductivity , but at very low temperatures , _
i.e. _ , at 0.36 and 0.5 k , @xmath3 shows upturn above some critical fields . since the zero - field thermal conductivity of bmo is purely phononic , the field - induced suppression of @xmath3 is clearly a result of scattering effect on phonons by some magnetic excitations . in this regard ,
the phonon scattering by paramagnetic moments , which are commonly existing as impurities or spin vacancies , is one of the unavoidable effects.@xcite however , the field dependencies of @xmath3 shown in fig .
3 are rather different from the standard one of paramagnetic moments scattering.@xcite the main origin for the @xmath47 behaviors is therefore the phonon scattering by magnon excitations .
the low - field quantum disordered phase has a finite spin gap , which can be weakened by the zeeman effect . at a fixed temperature , with increasing the magnetic field ,
the spin gap decreases and the number of low - energy magnons increases quickly , which can strongly scatter phonons and leads to a significant suppression of @xmath3 . a notable feature of @xmath47 isotherms is the upturn of @xmath3 in high magnetic field and at very low temperatures .
this increase of @xmath3 is obviously related to the field - induced af order since the transition fields are found to be coincided with the phase boundary of high - field ordered phase , _
i.e. _ , the magnon bec state .
there are two possible reasons for the increase of @xmath3 in the bec state .
first , the phonon scattering by magnons is weakened when the magnon bec occurs .
second , the magnons in the bec state have a positive effect on @xmath3 by transporting the heat directly .
if the magnons transport the energy , due to the anisotropy of the exchange and the low - dimensionality of the spin structure , the field dependence of @xmath3 should exhibit obvious anisotropy on the direction of heat current with the same field.@xcite however , it can be seen from fig .
3 that the upturn behaviors actually show a weak anisotropy on the direction of magnetic field rather than on that of the heat current .
this means that the upturn of thermal conductivity in the bec state is a result of the weakening of magnon - phonon scattering . in other words , although the magnon scattering on phonons are generally strong at high fields , it is more significant at the critical fields because of the strong spin fluctuations .
although the above discussions are quite clear in catching the main physics of the observed heat transport properties , it is useful to try a more quantitative analysis on the experimental data . first , since the phonons are the only one type of heat carriers , the zero - field @xmath41 data are likely to be fitted by a classical debye model of phonon thermal conductivity@xcite @xmath48 in which @xmath49 is the phonon frequency , @xmath50 is dimensionless , and @xmath51 is the phonon relaxation rate . the relaxation time is determined by @xmath52 which represent the phonon scattering by the grain boundary , the point defects , the phonon - phonon umklapp scattering , and the resonant scattering , respectively.@xcite the average sound velocity @xmath53 can be calculated from the debye temperature using the formula @xmath54 , where @xmath55 is the number density of atoms . for bmo
, however , the debye temperature can not be simply obtained from the phonon specific - heat data , as the data in fig . 1 indicate . for an estimation ,
we refer the specific - heat data of ba@xmath0v@xmath1o@xmath2,@xcite a nonmagnetic material having the same crystal structure as bmo .
the phonon specific - heat coefficient @xmath56 j / k@xmath7mol can be easily obtained from a @xmath32-fitting to the low-@xmath30 data.@xcite with this @xmath57 value , the debye temperature and the sound velocity of bmo are calculated to be 346 k and 2900 m / s , respectively .
the parameter @xmath58 describing the boundary scattering is the the averaged sample width of the samples .
the other parameters @xmath59 , @xmath60 and @xmath61 are free ones .
one difficulty of getting a precise calculation using eq . ( [ kappa ] ) is the description of @xmath62 . in bmo
, the magnetic excitations can strongly scatter phonons . considering the energy dispersion of magnons ,
it is very difficult to get a transparent formula for @xmath62 .
* for a simple analysis , one can neglect the magnon dispersion and assume that the resonant scattering of phonons occurs via a singlet - triplet excitations of dimerized states .
thus , the resonant scattering rate can be expressed as @xmath63 where @xmath64 is a free parameter while @xmath65 describes the difference of thermal populations of the excited triplet and the ground singlet states ( the contribution of a higher - level excitations of quintuplet states is negligibly small ) .
it is known that@xcite @xmath66 with @xmath67 the energy gap of magnetic spectrum , which is about 14.2 k from the specific - heat data .
data ( a ) and the @xmath68 data ( b ) with the calculation results using the debye model.,width=226 ] the zero - field thermal conductivity data are able to be fitted by using the above formula very well . as an example
, the zero - field data from fig . 2(a ) and the fitting curve are displayed in fig .
the best fitting parameters are @xmath59 = 1.3 @xmath28 10@xmath69 s@xmath29 , @xmath60 = 3.0 @xmath28 10@xmath70 k@xmath71s@xmath72 , @xmath61 = 6.0 , and @xmath64 = 1.6 @xmath28 10@xmath73 s@xmath71 .
second , the magnetic - field dependence of @xmath3 can also be calculated using the debye model .
the effect of magnetic field is to induce the zeeman splitting of the degenerate triplet states and a linear decrease of the energy gap with field , that is , @xmath74 the land factor was known to be @xmath75.@xcite a typical comparison between the @xmath47 data and the calculations is shown in fig .
although there is a qualitative agreement between the experimental data and the calculation , the quantitative difference is quite large . in particular , at low fields the experimental data show much stronger suppression than the calculated result .
it is expectable from the classical boltzmann distribution that the magnon excitations are negligibly weak at subkelvin temperatures when the spin gap is larger than 10 k. the experimental data suggest that the magnon excitations , which may be because of the quantum fluctuations , are much stronger than the thermal excitation produces .
another reason for the significant discrepancy is likely related to the simple assumption of the resonant phonon scattering by the two - level magnetic excitations . actually , the magnon dispersions of bmo are not very weak , which makes the scattering between phonons and magnetic excitations inelastic and much more complicated . a precise calculation based on the magnon - phonon scattering calls for the details of both the phonon spectra and the magnon dispersions.@xcite
it is useful to compare these results with those of dtn , in which the magnetic field along the @xmath5 axis ( the spin - chain direction ) can close the spin gap and drive the af phase transition . in dtn , the heat transport along the @xmath5 axis or perpendicular to it present remarkably anisotropic behaviors with increasing field , that is , @xmath27 show sharp peaks at @xmath17 while @xmath26 show dips at the same position.@xcite those demonstrated that magnons act as the heat carriers along the @xmath5 axis and the phonon scatterers along the @xmath4 plane , respectively.@xcite furthermore , the @xmath27 of dtn in the af state ( for @xmath76 ) keep increasing with lowering the temperature , also pointing to a potentially large heat transport in the magnon bec phase.@xcite the present @xmath47 behaviors of bmo for both @xmath26 and @xmath27 are very similar to those of @xmath68 in dtn , which suggests that the magnetic scattering on phonons are enhanced upon approaching the critical fields .
in contrast to the case of dtn , the magnetic excitations of bmo do not contribute to transporting heat substantially along either the @xmath4 plane or the @xmath5 axis .
therefore , whether the magnon bec state of the spin - gapped compounds can have strong ability of transporting heat is mainly dependent on such factors as the nature of the magnetic structure and the exchange anisotropy etc .
the field - induced af state or magnon bec is not the sufficient condition for observable magnon transport . .
the arrows indicate the critical temperatures where @xmath3 starts to increase.,width=321 ] the above are , however , not the whole story because there is only one transition in each @xmath47 isotherm for both @xmath77 and @xmath45 , whereas the former experiments revealed a single magnetic transition and two distinct transitions , respectively . for more precise description ,
we show in fig .
5 a series of @xmath41 curves for magnetic fields varying between 10 and 14 t. it is found that below 10 t , there is no any anomaly in the @xmath41 curves down to 0.3 k. above 10 t , a jump in @xmath41 shows up and becomes bigger with increasing field up to 14 t. from these data , we can obtain the transition points on the @xmath78 phase diagram and compare them with the boundaries of the magnetically ordered phases determined by heat capacity , magnetocaloric effect , and cantilever torque measurements,@xcite as shown in fig . 6
it is interesting that the jump - like anomalies of @xmath3 occur at the boundary of phase i , that is , the thermal conductivity increases at the phase transitions to phase i from either the low - field disordered state or the antiferromagnetically ordered phase ii .
this phenomenon has a good correspondence to the specific - heat data,@xcite which shows a sharp lambdalike peak and a much weaker and less divergent peak at the phase transitions from the disordered state or phase ii to phase i and the disordered state to phase ii , respectively .
it is likely that the phase transition associated with lambdalike specific - heat anomaly has more significant critical fluctuations that strongly scatter phonons .
curves from fig .
temperature - field phase diagram , obtained by heat capacity and magnetocaloric effect measurements from ref .
, is shown for comparison .
i and ii indicate two different phases of the triplet condensate.@xcite , width=264 ]
we study the heat transport of ba@xmath0mn@xmath1o@xmath2 single crystals at very low temperatures and in high magnetic fields to probe the field - induced magnetic phase transitions and the role of magnons in the transport properties . in zero field ,
the low-@xmath30 thermal conductivity shows a purely phononic transport behavior . in applied strong field ,
the spin gap is gradually diminished and the magnetic excitations are populated , which induces an enhanced magnetic scattering on phonons and an overall strong suppression of thermal conductivity with increasing field . moreover , at very low temperatures where the closure of spin gap results in a long - range ordered af state , the phonons are more significantly scattered at the phase transitions .
there is no evidence showing the sizeable magnetic heat transport in this magnon bec compound .
this work was supported by the chinese academy of sciences , the national natural science foundation of china , and the national basic research program of china ( grants no .
2009cb929502 and no .
2011cba00111 ) .
y. kohama , a. v. sologubenko , n. r. dilley , v. s. zapf , m. jaime , j. a. mydosh , a. paduan - filho , k. a. al - hassanieh , p. sengupta , s. gangadharaiah , a. l. chernyshev , and c. d. batista , phys .
* 106 * , 037203 ( 2011 ) .
|
we report the study on the low - temperature heat transport of ba@xmath0mn@xmath1o@xmath2 single crystal , a layered spin - dimer compound exhibiting the magnetic - field - induced magnetic order or the magnon bose - einstein condensation .
the thermal conductivities ( @xmath3 ) along both the @xmath4 plane and the @xmath5 axis show nearly isotropic dependence on magnetic field , that is , @xmath3 is strongly suppressed with increasing field , particularly at the critical fields of magnetic phase transitions .
these results indicate that the magnetic excitations play a role of scattering phonons and the scattering effect is enhanced when the magnetic field closes the gap in the spin spectrum .
in addition , the magnons in the bec state of this materials do not show notable ability of carrying heat .
|
atoms and molecules interacting with a strong laser field may be ionized by absorbing more photons than necessary .
this highly nonlinear phenomenon is called above - threshold ionization ( ati ) and has attracted increasing attention since its 1979 discovery by agostini and co - workers @xcite . recently
, impressive progress has been achieved in the study of ati .
for example , great emphasis has been placed on employing ati as a tool for the attosecond imaging of molecular orbitals @xcite .
imaging applications are based on the physical mechanism behind this phenomenon , in which the active electron ( i ) is released in the continuum , ( ii ) is accelerated by the field , and ( iii ) reaches the detector without returning or after rescattering with its parent molecule .
the former and the latter scenario are known as direct ati , and as rescattered or high - order ati ( hati ) , respectively .
quantum mechanically , this electron may be released along several pathways , and the corresponding transition amplitudes will interfere .
previous studies of ati have shown that for the strong - field ionization of molecules , there are two kinds of quantum interference displayed in the ati spectrum . in the first scenario , for a given energy ,
the electronic wave packet may reach the continuum at events occurring at different times in one cycle of laser field @xcite .
hence , the transition amplitudes associated with ionization along different orbits in the time domain will interfere .
this leads to a fringe structure in the ati spectrum @xcite , which is present regardless of whether the target is an atom or a molecule .
for that reason , we will refer to this type of interference as temporal interference " .
recently , ati spectra with unusual two - photon separation in the direction perpendicular to the laser polarization @xcite have been revealed due to the temporal interference . in the second scenario , due to the multi - core structure of molecules
, there may be electron emission from spatially separated centers , which results in a multi - slit like interference pattern in the ati spectrum . in the context of
linearly polarized fields and diatomic molecules , spatial - interference effects have been widely studied @xcite .
such studies range from simplified molecular models to exact three - dimensional time - dependent calculations . employing the molecular strong - field approximation ( sfa )
, a generalized spatial interference condition for diatomic molecules that accounts for the orbital geometry and @xmath0 mixing has been derived @xcite . for such fields , however , the dynamics are mainly confined along the field polarization axis .
elliptically polarized laser fields exhibit a nonvanishing electric - field component perpendicular to the major polarization axis , which affects the motion of the electron in the laser field .
hence , they have been proposed as a resource for controlling strong - field phenomena by varying the ellipticity .
for example , it is known that elliptical polarization has great significance as a tool to produce isolated attosecond pulses @xcite and to increase the contribution of long orbits in the hati process
@xcite . so far , the ellipticity dependence of ati has been widely studied since the mid 1990s .
most studies , however , focus on the influence of the field ellipticity on the angular dependence of the ionization rate of ati @xcite , while the imprints left on the ati spectrum by a nonvanishing driving - field ellipticity have received little attention . in an early paper @xcite ,
features related to the above - mentioned temporal interference have been identified experimentally for direct ati .
therein , it was found that the photoelectron yield as a function of the driving - field ellipticity behaves in two very distinct ways for hati and direct ati energy ranges .
while in the former case this yield decreases monotonically for larger driving field ellipticity , following the predictions of classical models , for direct ati the electron ellipticity distributions display several shoulder " structures .
these structures have been related to the quantum temporal interference between the two sets of orbits that contribute to ati in this energy region .
however , how the ellipticity influences the two orbits still remains unexplored to a great extent . in this work ,
we are interested in how the driving - field ellipticity influences the above - mentioned two kinds of interferences in the ati spectra of diatomic molecules , exemplified by n@xmath1 and ar@xmath1 . throughout
, we focus on direct ati , for which temporal interference effects are also prominent . employing the molecular sfa and saddle - point methods , we find that , in general , the temporal interference in the ati spectra will become blurred in the elliptically polarized field . for electron momenta along the major polarization axis , however , the temporal interference remains clear and the positions of the interference minima shift towards lower photoelectron energies with increasing ellipticity .
these features are well explained in terms of the ellipticity dependence of the initial electron velocity along each orbit , and we show that the downshift of the interference minima is responsible for the shoulder " structure observed in the ellipticity distributions of ati electrons in ref .
@xcite .
furthermore , for the spatial interference , clear interference minima which do not depend on the ellipticity are found for n@xmath2 if only @xmath3 or @xmath4 states are taken .
however , if @xmath0 mixing is introduced , sharp fringes are only present for linear polarization if the molecule is aligned perpendicular to the field axis . for all other cases ,
these minima are blurred even for linear polarization .
for ar@xmath2 , the ati spectra display several clear minima caused by the spatial interference .
the reasons for the differences between n@xmath1 and ar@xmath1 molecules are discussed in detail in terms of the shapes of their highest occupied molecular orbitals ( homo ) . in this view
, we propose a necessary condition to observe the spatial interference in the ati spectrum in experiments .
this paper is organized as follows . in sec .
ii , we provide a brief discussion of the model , including how to obtain the ati amplitude and two - center interference conditions .
subsequently , we present the ati spectra of the n@xmath2 and ar@xmath2 molecules in the elliptically polarized laser field , and discuss how quantum interference is altered by the driving field of non - vanishing ellipticity . finally , in sec .
iv our conclusions are given .
atomic units ( a.u . ) are used throughout unless otherwise indicated .
within the molecular sfa , the transition amplitude for direct ati from the initial bound state @xmath5 to a final volkov state with drift momentum @xmath6 is given by carla2002pra , becker2002advatmoloptphys , kfr @xmath7,\ ] ] where @xmath8^{2}d\tau + i_{p}t \label{action}\ ] ] is the semiclassical action .
( [ mpdir ] ) and ( [ action ] ) , @xmath9 gives the ionization potential , and @xmath10 and @xmath11 denote the laser electric field and the vector potential , respectively . for simplicity , we assume that all the molecular structure is contained in the ionization prefactor @xmath12this is the most widespread assumption when performing the sfa modeling of strong - field phenomena in molecular systems ( @xcite ; see also the reviews @xcite ) , and implies that the active electron leaves from the geometrical center of the molecule . in this work
, we employ the single - active orbital approximation , i.e. , we assume that the initial state @xmath13 is the homo .
this orbital is written as the linear combination of atomic orbitals ( lcao ) , and the core is assumed to be frozen .
this latter approximation is reasonable for molecules with heavy nuclei @xcite .
explicitly , @xmath14 , \]]where the wavefunctions @xmath15 give the atomic orbitals , @xmath16 is the internuclear distance , @xmath17 is the orbital quantum number , @xmath18 is the magnetic quantum number and @xmath19 for gerade ( g ) and @xmath20 for ungerade ( u ) orbital symmetry @xcite . for sufficiently high intensity and low frequency of the laser field , the temporal integrations in the amplitude ( [ mpdir ] )
can be mathematically evaluated by the saddle - point method with high accuracy @xcite , i.e. , we seek solutions such that the action ( [ action ] ) is stationary . for the specific case of an elliptically polarized field confined to the @xmath21 plane , the saddle - point equation to be solved reads as @xmath22^{2}}{2}+\frac{[p_{y}+a_{y}(t)]^{2}}{2}+i_{p}=0 , \label{saddle}\]]where @xmath23 and @xmath24 denote the final momentum in the @xmath25 direction and in the @xmath26 direction , respectively .
physically , eq . ( [ saddle ] ) ensures the conservation of energy at the ionization time @xmath27 . in terms of the solutions
@xmath28 of eq .
( saddle ) , the transition amplitude ( [ mpdir ] ) can be written as @xmath29 , \label{sfa_saddle}\]]where the prefactors @xmath30 are expected to vary much more slowly than the action at each saddle for the above - stated approximation to hold @xcite . for each final momentum @xmath6 , there are two solutions of eq .
( [ saddle ] ) within one period of the laser field , whose contributions will interfere . this interference is present regardless of whether the target has one or more centers , i.e. , for atoms and molecules , and has been extensively studied in the literature for linearly polarized driving fields @xcite . in this work ,
we employ the monochromatic elliptically polarized laser field @xmath31with vector potential @xmath32where @xmath33 is the laser ellipticity . in this case , all cycles are equivalent and the amplitude @xmath34 can be intuitively understood as the coherent superposition of the contributions of the two orbits , which may add constructively or destructively . for vanishing electron momenta , these orbits will start at two consecutive maxima of the driving field @xmath10 .
these times will approach each other as the momentum increases . throughout
, we will refer to the orbits starting in the first and second half cycle of the field ( [ et ] ) as orbit 1 and orbit 2 , respectively . for a discussion of these solutions for linearly polarized monochromatic fields
see @xcite .
due to the two - center structure of the diatomic molecule , one may expect electron emission from spatially separated centers in the molecule .
this results in minima and maxima in the ati spectra , which , in the present model , are implicit in the ionization dipole matrix element @xmath35 . using the length gauge with field - dressed molecular bound states dressedstates , the ionization prefactor ( [ prefactor ] ) can be rewritten as @xmath36 \mathbf{e}(t)\cdot i\partial _
{ \tilde{\mathbf{p}}}\psi _ { \alpha } ( \tilde{\mathbf{p } } ) \end{split}\ ] ] where @xmath37 is the fourier transform of the atomic orbitals @xmath15 given by eq .
( [ molecule ] ) with @xmath38 .
the two - center interference is related to the terms in the bracket in eq .
( [ prefactor2 ] ) .
this generalized interference condition accounts for @xmath0 mixing and the orbital geometry , and reduces to simpler interference conditions if only @xmath3 or @xmath4 states are present .
for example , the terms in bracket reduce to @xmath39 for @xmath3 states and gerade ( g ) symmetry .
thus , interference minima are present if @xmath40where @xmath41 denotes an integer number .
similarly , the interference minima are given by the condition @xmath42 for @xmath4 states and gerade symmetry .
if , on the other hand , the homo has ungerade ( u ) symmetry , the above - stated scenario is reversed , i.e. , the destructive interference condition for @xmath4 states is given by eq .
( [ sstate ] ) , while eq .
( [ pstate ] ) holds for @xmath3 states . in this paper , the molecular axis
is kept in the plane @xmath21 spanned by the driving - field ellipticity .
the frame of reference ( @xmath43,@xmath44,@xmath45 ) of the molecule is rotated by an angle @xmath46 in this plane , defined with regard to the major polarization axis of the laser field .
this means that @xmath47 , @xmath48 , and @xmath49 .
the wave functions @xmath15 in eq .
( [ molecule ] ) will be approximated by a gaussian basis set @xmath50where the coefficients @xmath51 , @xmath52 , @xmath53 and the exponents @xmath54 are extracted from the quantum chemistry code gamess - uk @xcite . throughout
, we employ a 6 - 31 g basis so that only @xmath3 and @xmath4 states are considered . for a gaussian basis set
, the wave function @xmath55 in momentum space can be written as @xmath56 usually , the explicit formula of @xmath57 is complicated . however , if only @xmath3 and @xmath4 states are used in the orbital construction , @xmath57 is simplified as @xmath58,\ ] ] where @xmath59 and the subscript @xmath60 indicates the orientation of the @xmath4 states employed to construct the orbital .
for instance , a @xmath61 orbital would require @xmath4 states along the @xmath43 axis , i.e. , @xmath62 , whereas a @xmath63 orbital would require @xmath4 states along the @xmath44 or @xmath45 axis , i.e. , @xmath64 or @xmath65 . in this work , we will study the n@xmath1 and ar@xmath1 molecules , both of which have @xmath61 homos . in combination with eqs .
( [ p_gauess ] ) and ( [ saddle ] ) , the prefactor in eq .
( [ prefactor2 ] ) can be specified as @xmath66 \mathcal{d}_{s}+ \\ & i \sin(\frac{\mathbf{p}\cdot \textbf{r}}{2})\sum_{\alpha(p),v } c_{\alpha } b_{v } \ { [ ( \tilde{p}_{x}^{2}-2\zeta_v ) \cos\eta + \tilde{p}_{x}\tilde{p}_{y}\sin \eta]\times \\ & e_{x}(t ) + [ ( \tilde{p}_{y}^{2}-2\zeta_v ) \sin\eta + \tilde{p}_{x}\tilde{p}_{y}\cos \eta ] e_{y}(t ) \ } \mathcal{d}_{p } \end{split}\ ] ] where @xmath67 and @xmath68 ) .
note that , in eq .
( [ prefactor3 ] ) , the momentum components relate to the electric field frame of reference .
in the results that follow , we will analyze how the above - mentioned temporal and spatial interference are influenced by a non - vanishing driving field ellipticity .
this analysis will be performed in terms of electron orbits . for the temporal interference
, we will focus on the n@xmath1 molecule , which has been widely studied in the literature , both theoretically and experimentally .
for all parameters employed in sec .
[ temporal ] , the structure caused by electron emission at separate centers gets washed out , so that temporal interference may be clearly assessed . in our studies of spatial interference effects ( sec .
[ spatial ] ) , we consider the n@xmath1 and ar@xmath1 molecules .
such molecules have very different homo geometries and equilibrium internuclear distances . molecule , coded in false colors in the ( @xmath69 , @xmath70 ) plane and plotted in a logarithmic scale . in panels ( a ) and ( d ) , both orbits 1 and 2 were included , for @xmath71 and @xmath72 , respectively , while in panels ( b ) , ( c ) , ( e ) and ( f ) the contributions of individual orbits were taken . in panels ( b ) and ( c ) , the field has been chosen to be linearly polarized ( @xmath71 ) , while in panels ( e ) and ( f ) the ellipticity is @xmath72 .
the major axis of the polarization ellipse is along the molecular axis , i.e. , for alignment angle @xmath73 .
the laser intensity and wavelength are @xmath74 w/@xmath75 and @xmath76800 nm , respectively . ] and the electric field @xmath77 , with @xmath78 [ panels(a ) and ( b ) , respectively ] .
the two points @xmath79 and @xmath80 denote the times associated to the two ionization events @xmath81 and @xmath82 that occur within a field cycle , for an electron with the same final momentum @xmath6 .
these events are indicated by the colored circles in the figure . for simplicity
, we have defined the time related to a field crossing @xmath83 for the electric - field component along the major polarization axis as @xmath84 .
this time also gives a maximum of the larger vector potential component @xmath85 .
note that @xmath86 as both events are equidistant in time from @xmath84 . ]
figure [ fig1 ] shows the ati spectra of the n@xmath1 molecule , coded in false colors in the ( @xmath69 , @xmath87 ) plane , where @xmath88 denotes the angle between the internuclear distance and the drift momentum , and @xmath89 is the kinetic photoelectron energy upon reaching the detector , in units of the ponderomotive potential @xmath90 .
we assume that the major axis of the polarization ellipse is along the molecular axis , i.e. , that the alignment angle @xmath91 is vanishing . for linear polarization
[ see fig .
[ fig1](a ) ] , many clear minima can be observed in the spectra .
these minima are due to the destructive interference between orbit 1 and orbit 2 .
to illustrate this more clearly , we calculate the ati transition probability density @xmath92 employing only individual orbits , for the same parameters as in fig .
[ fig1](a ) .
these results are displayed in figs .
[ fig1](b ) and ( c ) , and unambiguously show no minima . with increasing ellipticity ,
the interference minima become blurred and , for most angles @xmath93 , completely disappear as the ellipticity reaches @xmath78 , as shown in fig .
[ fig1](d ) .
in fact , only the minima along and in the vicinity of the major polarization axis ( @xmath94 and @xmath95 ) remain visible for this ellipticity .
sharp minima are only present exactly along the axis . , the saddle - point eq .
( [ saddle ] ) exhibits two relevant solutions in the complex time plane for @xmath96 , whose imaginary parts are identical and which merge into a single solution around @xmath97 .
it is only for these larger ellipticities that the fringes will vanish for this specific angle . ]
this behavior is due to the fact that , in general , a non - vanishing driving - field ellipticity affects the initial electron velocity along orbit 1 and orbit 2 unequally .
the initial velocity of the electron at the ionization time @xmath27 is @xmath98 , and thus the velocity along the minor polarization axis is given by @xmath99 . in fig .
[ vector](a ) , we present the time evolution of the vector potential @xmath11 with ellipticity @xmath100 . the two times @xmath79 and @xmath80 represent the ionization times related to orbit 1 and 2 , respectively , and are equidistant from the electric field crossing @xmath83 [ see fig .
[ vector](b ) ] .
the figure clearly shows that the directions of the vector potential @xmath101 are opposite at the two ionization times .
thus if the electron leaves along orbit 1 , the absolute value of its initial velocity @xmath102 should be larger than if it leaves along orbit 2 .
this holds except for linear polarization , i.e. , if @xmath103 , or for electron momenta along the major polarization axis , i.e. , for @xmath94 or @xmath95 .
the ionization rate at a specific ionization time @xmath27 is evaluated by @xmath104 @xcite , where @xmath105 , \label{w0}\]]@xmath106 , \label{w1}\]]and @xmath107 is the initial velocity .
hence , the larger the initial velocity @xmath108 is , the lower the ionization rate at that ionization time will be .
this implies that , for elliptical polarization , ionization along orbit 2 will become larger than ionization along orbit 1 .
consequently , the destructive interference between two electron orbits will disappear in the ati spectra .
here we assume that the transverse velocity of the electron is @xmath109 . if @xmath110 the above - stated results are reversed .
this is in agreement with figs .
[ fig1](e ) and ( f ) , in which we present the individual contributions to the ati spectra from orbit 1 and orbit 2 , respectively , for ellipticity @xmath100 . in general , the probability density @xmath92 associated with orbit 1 is much lower than that related to orbit 2 .
this is in stark contrast to what is observed for linear polarization , for which the two probability densities are identical [ see figs .
[ fig1](b ) and ( c ) ] .
however , for drift momentum @xmath111 along the major polarization axis ( @xmath94 and @xmath95 ) , the ati probability densities are the same , because the two orbits exhibit the same initial perpendicular velocity @xmath112 .
therefore , the coherent superposition of the corresponding transition amplitudes will lead to clear interference minima in the ati spectra , as observed in fig .
[ fig1](d ) .
molecule along the major polarization axis as a function of the alignment angle @xmath46 , coded in false colors in the ( @xmath69 , @xmath46 ) plane .
ellipticity @xmath113 for ( a ) and @xmath78 for ( b )
. the parameters of the laser field are the same as in fig .
[ fig1 ] . ] ) for parallel alignment @xmath73 and momentum along the major polarization axis ( @xmath114 ) with @xmath71 ( dashed lines ) and @xmath72 ( full lines ) : action ( black ) , prefactor ( red ) , and total transition amplitudes ( green ) .
the gray arrows indicate the values of the energy when the total phase difference is equal to @xmath115 and @xmath116 for @xmath113 and @xmath78 .
( b ) saddle points associated with the solutions of eq .
( [ saddle ] ) for the two orbits as functions of the ellipticity from @xmath113 to @xmath78 , for parallel alignment , momenta along the major polarization axis , and kinetic energies @xmath117 ( black full lines ) , @xmath118 ( red dashed lines ) and @xmath119 ( blue dotted lines ) .
the gray arrows indicate the real values of the saddle points for @xmath113 and @xmath78 . ]
next we will further study the interference minima in the ati spectra along the major polarization axis . in fig . [ moving ]
, we calculate the ati spectra of the n@xmath1 molecule with @xmath114 as a function of the alignment angle @xmath46 .
the most eye - catching features in the spectra are the clear minima due to the destructive interference between the two electron orbits . with increasing ellipticity ,
the ati spectra seem to be squeezed and the positions of the minima are moving down to lower energies .
for example , in fig .
[ moving](a ) the position of one of the minima is about @xmath120 for @xmath73 and @xmath113 , but it becomes @xmath121 in fig .
[ moving](b ) for @xmath78 .
furthermore , fig .
[ moving ] shows that the downshift of the minima with the field ellipticity is much more dramatic for higher photoelectron energies than in the low - energy region . in order to understand the behavior of the minima in fig . [ moving ] , we calculate the phase difference between the two orbits in eq .
( [ sfa_saddle ] ) , which determines the positions of the interference minima in the ati spectra . in fig .
[ action](a ) , we exhibit the phase difference between the amplitudes @xmath122 and @xmath123 , the actions @xmath124 and @xmath125 , and the prefactors @xmath126 and @xmath127 associated with the two orbits for parallel alignment @xmath73 , and field polarizations @xmath71 and @xmath128 ( dashed and full lines , respectively ) .
the figure shows that the phase difference between @xmath124 and @xmath125 ( black lines ) is smaller for @xmath78 than for @xmath113 .
this is caused by the real parts of two saddles @xmath129 and @xmath130 approaching each other for elliptical polarization , and @xmath130 are the same for a fixed polarization , the change in the real parts are responsible for the shifts of minima in the patterns . ]
as illustrated in fig .
[ action](b ) .
moreover , the approaching of the two saddles becomes more significant in the high - energy than in the low - energy region of the photoelectron spectra [ see the gray arrows in fig .
[ action](b ) ] . therefore
, the decrease of the phase difference associated with the action is much more dramatic for higher photoelectron energies than in the low - energy region . for the prefactor ,
the phase difference changes much less markedly with regard to the ellipticity , as shown in the red lines in fig .
[ action](a ) .
this is in agreement with the assumption of the steepest descent method that the prefactor varies more slowly with time than the action @xcite .
thus , the change in the total phase difference [ green lines in fig .
[ action](a ) ] with the ellipticity is mainly determined by the action .
when the difference in the total phase is equal to @xmath131 , destructive interference between the two orbits will occur . in fig .
[ action](a ) , the positions of the total phase difference with @xmath132 and @xmath133 are marked with arrows .
the two positions are about @xmath120 and @xmath134 for linear polarization and move down to about @xmath121 and @xmath135 , respectively , for elliptical polarization .
moreover , the downshift of the minima in the high - energy region is more significant than that in the low - energy region .
these results are consistent with the behaviors of the interference minima observed in fig . [ moving ] . furthermore , the behavior of the real parts of the saddle points in fig .
[ action](b ) can be explained in terms of the ellipticity dependence of the initial ionization velocity @xmath108 . in the classical limit
, the real parts @xmath136 and @xmath137 are associated with the classical ionization times @xmath129 and @xmath130 of an electron in the field ( [ et ] ) , obtained if @xmath138 . in a linearly polarized laser field , if an electron is released with drift momentum @xmath6 along the major polarization axis , it will reach the continuum with the vanishing initial velocity @xmath139 . in an elliptically polarized field , however , the electron should have a nonvanishing initial velocity in order to compensate the motion induced by the small component of the field . in the classical limit
, the total initial velocity @xmath140 of the electron is perpendicular to the laser electric field @xmath10 at the corresponding ionization time @xmath27 @xcite , i.e. , @xmath141 .
the condition @xmath142 , @xmath143 requires that @xmath144 and @xmath145 for the parameters employed in this paper .
since @xmath146 and @xmath147 for both @xmath129 and @xmath130 [ see fig .
[ vector](a ) ] , an increase in the value of @xmath148 with the driving - field ellipticity requires that @xmath129 increases in order to approach the time for which @xmath85 is maximal , i.e. , @xmath84 [ see fig . [ vector ] ] .
moreover , for the photoelectrons in the high energy region , the ionization time is closer to @xmath84 @xcite .
hence , a larger initial velocity @xmath149 will be required to satisfy the condition @xmath150 .
this results in a larger increase of the ionization time .
the same argument applies to orbit 2 , with the difference that , because this orbit is located after the electric - field crossing [ see fig .
[ vector](b ) ] , @xmath130 should decrease . as a direct consequence , for increasing ellipticity the two ionization times should approach each other , and the approaching of the two times becomes more significant for higher photoelectron energies than in the low - energy region .
these results are in good agreement with the two saddle - point solutions as functions of the driving - field ellipticity shown in fig .
[ action](b ) ( or see fig . 4 in ref .
@xcite ) .
it is worth mentioning that , because the interference minima in fig .
[ moving ] move down to lower energy with increasing ellipticity , for fixed photoelectron energy there will be fluctuations in the ati spectrum .
this interesting phenomenon has been observed in experiments @xcite and was called the shoulder " structure in the ellipticity distribution of the ati electrons . in our work , the underlying mechanism of the shoulder " structure is explained in terms of the ellipticity dependence of the initial ionization velocity .
molecule computed using only the s states [ panels ( a ) and ( c ) ] and the p states [ panels(b ) and ( d ) ] from its homo , coded in false colors in the ( @xmath69 , @xmath151 ) plane . the major axis of the polarization ellipse is along the molecular axis , @xmath152 .
ellipticity @xmath71 for ( a ) and ( c ) and @xmath72 for ( b ) and ( d ) . in all panels ,
the two - center interference minima are indicated by the white curves in the figure .
the remaining field and molecular parameters are the same as in fig .
[ fig1 ] . ]
finally , we will study the spatial interference due to electron emission at the two spatially separated centers in n@xmath1 and ar@xmath1 .
the former molecule has a @xmath153 homo with a high degree of @xmath0 mixing . for the latter molecule ,
the homo is a @xmath154 orbital , in which the @xmath4 states dominate .
furthermore , ar@xmath1 exhibits a very large equilibrium internuclear distance ( @xmath155 a.u . ) , so that , according to eq . ( [ prefactor2 ] ) , several two center minima are expected to be present in the momentum range of interest . in fig .
[ ati_sp ] , we display the ati spectra of n@xmath1 , if only the contributions from the @xmath3 or @xmath4 states to the homo are taken .
these spectra are coded in false colors in the ( @xmath69 , @xmath87 ) plane .
the major axis of the polarization ellipse is chosen to be parallel to the molecular axis , i.e. , at the alignment angle @xmath73 .
[ ati_sp](a ) shows that for the @xmath3-state contributions and linear polarization , apart from the minima caused by the temporal interference , there are other two minima , whose positions satisfy the two - center interference condition , eq .
( [ sstate ] ) , shown by the white lines in the figure . with increasing ellipticity [ see fig .
[ ati_sp](c ) ] , the minima from the temporal interference are blurred as discussed in the previous section , but the minima from the two - center interference are still very clear .
that is because the two - center interference condition , eq .
( [ sstate ] ) , is given for field - dressed initial bound states @xcite .
in this case , it is independent on the ellipticity .
similarly , for the @xmath4-state contributions , the two - center interference minima appear in the middle of the each panel in figs .
[ ati_sp](b ) and ( d ) , because the corresponding electron momentum is perpendicular to the molecular axis , i.e. , @xmath156 .
hence , the interference condition , eq .
( [ pstate ] ) , always holds .
however , when we consider the full @xmath153 homo of the n@xmath1 molecule and include @xmath0 mixing , the minima from the two - center interference can not be observed in the ati spectra .
this holds for linearly and elliptically polarized fields , as shown in figs .
[ fig1](a ) and ( d ) , and even if only the contributions of individual orbits are taken , as illustrated in figs .
[ fig1](c ) , ( b ) , ( e ) , and ( f ) .
there are two main reasons for it .
first , if @xmath0 mixing is included , the interference condition also depends on the sum of @xmath157 in eq .
( [ prefactor2 ] ) , where @xmath38 .
this influence is expected to be considerable for @xmath158 , as its @xmath159 homo has a high degree of @xmath0 mixing .
second , the ionization time @xmath27 has a positive imaginary part , which can be related to the width of the barrier through which the electron must initially tunnel @xcite .
thus , the sum in eq .
( [ prefactor2 ] ) is usually not purely real or imaginary , and therefore , the absolute value of eq .
( [ prefactor2 ] ) will never completely vanish , i.e. , no clear minima will be observed in the ati spectra .
molecule , coded in false colors in the ( @xmath69 , @xmath70 ) plane .
the major axis of the polarization ellipse is perpendicular to the molecular axis , @xmath160 .
ellipticity @xmath71 for ( a ) and @xmath72 for ( b ) .
( c ) and ( d ) ati spectra obtained by employing only individual orbits in the ( @xmath161 , @xmath151 ) for @xmath71 .
the white curves indicate the minima caused by the two - center interference . ]
nevertheless , there is a specific case in which clear spatial interference fringes are present in the full ati spectra of n@xmath2 , if @xmath0 mixing is included .
[ ati_sp_90](a ) shows the spectra for linear polarization with the alignment angle @xmath160 .
apart from the minima from the temporal interference , there are two other minima . in order to show them more clearly
, we calculate the ati transition probability density @xmath92 by employing only individual orbits to remove the pattern associated with the temporal interference . figs .
[ ati_sp_90](c ) and ( d ) display the ati spectra without the temporal interference , where the clear minima from the spatial interference are observed .
the reason for the appearance of the spatial interference in this specific case is that for the perpendicular alignment @xmath162 and for linear polarization , the ionization prefactor in eq .
( [ prefactor3 ] ) can be simplified as @xmath163 where the momentum @xmath164 in the @xmath165 direction is real .
thus , @xmath166 can be factorized in both the first and the second terms in eq .
( [ prefactor4 ] ) , and the remaining terms are time - independent and purely real . using eq .
( [ prefactor4 ] ) , we obtain the minima from the destructive spatial interference in the ati spectra , shown by the white curves in figs .
[ ati_sp_90](a ) , ( c ) , and ( d ) , which agree well with what is observed in the spectra . on the other hand , for elliptical polarization , the imaginary parts in the two terms in eq .
( [ prefactor3 ] ) are different .
hence , they can not be factorized and the absolute value of eq .
( [ prefactor2 ] ) will never vanish .
this is also the case if the field is linearly polarized , but @xmath167 , as seen from eq .
( [ prefactor4 ] ) .
[ ati_sp_90](b ) shows that as the ellipticity increases to @xmath78 , the minima from both the spatial and temporal interference in the spectra fade .
molecule , with internuclear distance @xmath155 a.u .
, coded in false colors in the ( @xmath69 , @xmath151 ) plane .
the major axis of the polarization ellipse is along the molecular axis , @xmath73 . the results in panels ( a ) and ( b ) have been computed for ellipticity @xmath71 and @xmath72 , respectively .
the minima caused by the two - center interference are indicated by white curves .
the remaining field parameters are the same as in fig .
[ fig1 ] . ] in contrast , the ati spectra of the ar@xmath1 molecule present clear minima from the two - center interference , as seen in fig .
[ ar2](a ) .
in addition , fig .
[ ar2](b ) shows that , with increasing ellipticity , most minima from the temporal interference are blurred .
however , the spatial minima remain unaltered .
the critical reason for the difference between ar@xmath2 and n@xmath2 molecules is that , in the @xmath154 homo of ar@xmath1 , the contributions of the @xmath4 states dominate and the contributions of the @xmath3 states are practically vanishing for the internuclear separation in this work .
hence , the corresponding two - center interference condition is given by eq .
( [ sstate ] ) , which holds for _ pure _ @xmath4 states and ungerade symmetry .
this condition , shown by the white lines in fig .
[ ar2 ] , is independent of the driving field and is consistent with the interference minima in the simulation .
furthermore , since ar@xmath2 has a very large equilibrium internuclear distance , @xmath155 a.u .
, several two - center minima are observed in the momentum range of interest according to eq .
( [ sstate ] ) .
in summary , we theoretically study temporal and spatial interference in the direct ati spectra of aligned diatomic molecules in an elliptically polarized laser field .
temporal interference relates to ionization events occurring at different times within a field cycle , and spatial interference is associated with electron ionization at spatially separated centers in the molecule . throughout , we have employed the molecular sfa , saddle - point methods and field - dressed bound states .
we have also assumed that only the homo is active .
with increasing ellipticity , there will be a blurring of the temporal - interference patterns , except when the photoelectron momenta @xmath111 are parallel to the major polarization axis .
this blurring is caused by the fact that a nonvanishing driving field ellipticity affects the electron velocities unequally for the two ionization events occurring within each field cycle .
for linear polarization , these events are equally probable . in order , however , to compensate the motion induced by the small component of the elliptically polarized field , it is necessary that , for one of the orbits , the electron be released in the continuum with a larger velocity .
this will lead to a suppression of the corresponding ionization probability , and hence to a disappearance of the temporal interference patterns .
nevertheless , if the electron is released along the major polarization axis , the absolute value of the initial velocity will be the same for the two ionization events .
hence , the temporal - interference minima in the ati spectrum will remain clear . for these specific parameters ,
we have analyzed in detail how a non - vanishing ellipticity shifts these minima towards lower energies , and have traced them to a decrease in the phase difference between the transition amplitudes related to both orbits .
these results are in agreement with the studies performed in @xcite .
our results also show that , in general , the spatial , two center interference is much more robust with regard to the electric field parameters than the temporal interference , as long as the homo does not exhibit a high degree of @xmath0 mixing . as a consequence of @xmath0 mixing , however , or of the temporal dependence of the ionization prefactor eq .
( [ prefactor2 ] ) , the interference patterns become blurred , even for linear polarization . only for the very specific case of linear polarization and a molecule aligned perpendicular to the field direction
will sharp fringes be observed .
this is exemplified by computations in n@xmath2 . on the other hand ,
if either @xmath3 or @xmath4 states are dominant in the homo , the patterns will be very clear over a large parameter range and practically independent of the driving - field ellipticity .
this is shown for the ati spectra of ar@xmath1 , which display clear spatial - interference patterns due to fact that the @xmath4-state contributions are dominant .
several minima exist due to the large equilibrium internuclear distance .
therefore , one can trace all the interference patterns observed , or the absence thereof , to the dynamics of the electronic wave packet and the velocity with which it is released in the continuum .
these patterns can be controlled by an appropriate choice of the the field and molecular parameters .
concrete examples provided in this work include the laser - field polarization , the molecular targets , the angle between the electron momentum and the major polarization axis , and the molecular orientation with regard to the field .
this control is related to the energy position and to the sharpness of the patterns .
we also identify a parameter region , for which clear interference fringes can be easily observed .
we expect this information to be useful for future experiments .
we are thankful to b. b. augstein for his help with gamess - uk .
this work has been funded by the uk epsrc ( grant ep / j019240/1 ) , nnsf of china ( grant no .
11204356 ) , and the cas overseas study and training program .
v. i. usachenko and shih - i .
chu , phys .
a * 71 * , 063410 ( 2005 ) .
m. lein , j. phys .
b * 40 * , r135 ( 2007 ) .
b. b. augstein and c. figueira de morisson faria , modern phys .
b * 26 * , 1130002 ( 2011 ) . c. b. madsen and l. b. madsen , phys .
a * 74 * , 023403 ( 2006 ) gamess - uk is a package of _ ab initio _ programs .
see http://www.cfs.dl.ac.uk/gamess-uk/index.shtml , m. f. guest , i. j. bush , h. j. j. van dam , p. sherwood , j. m. h. thomas , j. h. van lenthe , r. w. a. havenith , j. kendrick , mol .
phys . * 103 * , 719 ( 2005 ) .
n. bleistein and r. a handelsman , _ asymptotic expansions of integrals " _
( dover , new york , 1975 ) r. kopold , phd thesis , technical university munich , chapt . 5 ( 2001 ) .
|
we investigate direct above - threshold ionization in diatomic molecules , with particular emphasis on how quantum interference is altered by a driving field of non - vanishing ellipticity
. this interference may be either temporal , i.e. , related to ionization events occurring at different times , or spatial , i.e. , related to the electron emission at different centers in the molecule . employing the strong - field approximation and saddle - point methods , we find that , in general , for non - vanishing ellipticity , there will be a blurring of the temporal and spatial interference patterns .
the former blurring is caused by the electron velocity component perpendicular to the major polarization axis , while spatial interference is washed out as a consequence either of @xmath0 mixing , or of the temporal dependence of the ionization prefactor .
both types of interference are analyzed in detail in terms of electron trajectories , and specific conditions for which sharp fringes occur are provided .
|
recent advances in fabrication techniques involving molecular beam epitaxy and lithography have now made it possible to make narrow gaas - based quasi - one dimensional ( 1d ) electronic systems with lateral dimensions of the order of the bohr radius @xcite . in these so - called quantum wire structures , the motion of charge carriers is confined in two transverse directions but is essentially free ( in the effective mass sense ) in the longitudinal direction .
study of collective modes in reduced dimensional electron systems in semiconductor nanostructures is a subject of growing experimental and theoretical interest .
experimentally , far - infrared optical spectroscopy@xcite and resonant inelastic light scattering spectroscopy@xcite have been used to study quasi-1d elementary electronic excitations .
several theoretical studies , mostly based on the random phase approximation ( rpa ) , have been reported on the energy dispersion of elementary excitations in semiconductor quantum wires .
the measured 1d intrasubband plasmon dispersion agrees remarkably well with the rpa predictions @xcite .
the quantitative agreement between rpa predictions and the experimental results @xcite was later explained by the fact @xcite that the calculated rpa plasmon dispersion and the tomonaga - lutinger theory for the collective charge density excitations of the 1d electron system are equivalent at long wavelengths by virtue of the vanishing of vertex corrections @xcite in the irreducible polarizability of the 1d electron gas .
the two - component quasi-1d systems which are composed of electrons and holes have been generated in a wide variety of semiconductor quantum wires by optical pumping .
spatially separated two - component quasi-1d systems can be produced by additional lateral confinement to produce a double quantum wire structure in gaas
ga@xmath4al@xmath5as double quantum well systems .
the collective modes of the two - component plasma also play important roles in the many - body physics of carriers such as screening of the coulomb interaction potential and in coulomb drag problem . in this paper
, we calculate the dispersion and the spectral weight of longitudinal collective modes ( with and without spatial separation ) in quasi-1d two - component plasmas as in semiconductor double quantum wire structures or in photoexcited homogeneous 1d electron - hole plasmas ( ehp ) .
our theory is based mostly on the two - component rpa ( which should be well - valid in 1d systems ) ; in the 1d ehp case we go beyond rpa to include local field corrections to assess the validity of 1d rpa .
the collective modes of the two - component plasma have been widely studied in higher dimension @xcite , and recently , numerically in quasi-1d ehp @xcite .
it is well - known that a two - component plasma has two branches of longitudinal collective excitation spectra called the optical plasmon ( op ) and the acoustic plasmon ( ap ) , where the density fluctuations in each component oscillate in - phase ( op ) and out - of - phase ( ap ) respectively relative to each other , assuming the two charge components to be the same .
as expected , we find two plasma modes in the two - component 1d system ; the in - phase op with a typical 1d plasmon dispersion with the frequency proportional to @xmath0 at long wavelengths and the out - of - phase ap with its frequency linear in @xmath2 as @xmath1 , where @xmath2 is the 1d wave vector and @xmath3 is a characteristic 1d confinement size ( i.e. , roughly the wire width ) . unlike 2d and 3d systems , where the electron - hole pair continua have only an upper boundary , @xmath6 , the 1d pair continuum has also a low energy gap under the lower boundary @xmath7 , implying no low energy single particle excitations ( spe ) are allowed in 1d .
this gap , the non - existence of low energy spe in 1d fermi systems , arises from severe phase space restrictions ( pauli exclusion principle ) imposed by 1d energy - momentum conservation , and leads to the possibility of the existence of an undamped plasmon mode in multi - component 1d systems@xcite .
we find a low energy undamped ap mode in the double - quantum - wire system for arbitrary interwire separations ( including the zero separation case ) .
in general , in higher dimensions , the collective mode spectral weight is mostly carried by the op , which makes the experimental observation of the ap mode in higher dimensions rather difficult .
in contrast , we find that in quasi-1d systems the spectral weight of the undamped ap mode is comparable to that of the op mode and the spectral weight of the ap mode increases as the spatial separation increases , which is similar to the corresponding 2d situation
. the greater relative spectral weight of the ap mode in 1d is directly related to the non - existence of low energy 1d spe .
the plasma modes in 1d systems have vanishing frequencies at long wavelengths , and in general their energies at experimentally accessible wave vectors are small ( @xmath8 mev ) .
their observability , therefore , depends rather crucially on the absence of appreciable damping or broadening in the system .
while the absence of long wavelength landau damping ( to spe ) makes it feasible to discuss the observability of the ap mode , one must also have very small impurity broadening in order for the plasmon modes to be observable .
impurity scattering causes the carrier motion to be diffusive assuming the scattering to be weak . in principle , any impurity disorder , no matter how weak , localizes all single electron 1d states ; for weak disorder , however , the localization length is larger than the system size and electron dynamics could be considered to be diffusive .
the diffusive nature of the electronic dynamics strongly affects the plasmon dispersion in the long wavelength limit because the plasma frequency is vanishingly small at long wavelengths .
the 1d plasmons therefore become overdamped due to impurity scattering induced level broadening at sufficiently long wavelengths .
this plasmon level broadening may actually play a significant role in determining the physical characteristics in quantum wires .
the collisional damping of the plasmon modes due to impurity scattering occurs in addition to the landau damping which arises from the decay of the plasmon to spe electron - hole pair excitations .
( as emphasized above , while landau damping is suppressed in 1d , impurity broadening is always present unless the system is perfectly pure . ) in the single - component plasma , it is known that , as the impurity scattering increases , the collisional damping becomes strong and the plasmon mode becomes overdamped and disappears at small @xmath2 in 1d and 2d systems ( by contrast , this is a small effect in 3d since the plasmon energy is finite at zero wave vector ) @xcite .
inclusion of impurity scattering in 1d and 2d overdamps the plasmon below a critical wave vector @xcite . this disappearance of the plasmon spectral weight at small @xmath2 has important consequences for the existence and characteristics of the fermi surface in a one dimensional electron gas @xcite . in this paper
we investigate the plasmon damping of the two - component 1d plasma in the presence of impurity scattering .
( as noted above any finite impurity disorder leads to anderson localization of 1d one electron states , and therefore the 1d electron dynamics in the presence of impurity disorder is exponentially localized , _ not _ diffusive this is , however , a purely academic point of little relevance to high quality gaas quantum wire systems with low disorder , where the localization lengths are many microns long and are in fact longer than the 1d sample lengths @xcite ; we are therefore justified in treating the effective electron dynamics as being diffusive , neglecting the anderson localization effects in high quality quantum wires of relevance to plasmon experiments . )
we also discuss the effect of interwire electron tunneling on the bi - wire collective mode spectra .
the most important qualitative feature of the plasmon dispersion in the presence of interwire tunneling is that , even though the in - phase op mode @xmath9 depends weakly on the tunneling , the out - of - phase ap mode @xmath10 , which is purely acoustic in the absence of tunneling ( @xmath11 ) , develops a plasmon gap at @xmath12 in the presence of nonzero tunneling ; the plasmon gap in the ap depends nontrivially on the 1d electron density @xmath13 , the interwire tunneling amplitude @xmath14 , and the interwire distance @xmath15 .
tunneling , therefore , effectively changes the ap ( @xmath11 ) into an optical mode with a long wavelength gap .
we use the rpa for most of our calculations . as emphasized
, rpa is a very good approximation for the collective mode dispersion in 1d quantum wire structures by virtue of the essential vanishing of all vertex corrections to the 1d polarizability function .
we consider the zero temperature case and assume that our quasi-1d system has an infinite square - well confinement with a finite width @xmath3 in the y - direction and a zero width in the z - direction .
it is easy to include a finite width in the @xmath16 direction but our results do not change qualitatively . in typical quantum wires , the wire width along the growth direction ( z - direction ) is much smaller ( by an order of magnitude ) than that in the lateral direction ( y - direction ) , justifying our 2d finite width model for the 1d wire . in the finite - width 2d model we use a square
well confining potential with infinite barriers at @xmath17 and @xmath18 , with the electron dynamics along the @xmath19-axis being free .
confinement of the electrons in the @xmath20 and @xmath16 directions leads to the quantization of energy levels into different 1d subbands , and we assume the 1d extreme quantum limit where only the lowest subband is occupied by carriers , and all the 1d higher subbands are neglected .
we study the plasmon modes of the spatially separated two - component system by taking two identical parallel quantum wires separated by a distance @xmath15 .
we include results for the case @xmath21 because it applies directly to the photoexcited 1d ehp where electrons and holes are not spatially separated .
we consider two different spatial configurations for our two - component bi - wire structure : two 1d wires with a spatial separation @xmath15 being parallel to each other aligned in the ( 1 ) @xmath22 plane ( separated along the @xmath20-axis ) and ( 2 ) @xmath23 plane ( separated along the @xmath16-axis ) .
since the difference in our calculated results between these two configurations is within one percent , we only give the results for the configuration ( 1 ) .
it is well - known that 1d electron systems are fundamentally different from their higher dimensional counterparts because 1d interacting electrons form a luttinger liquid , not a fermi liquid , and any electron - electron interaction , no matter how weak , destroys the fermi surface , i.e. , the discontinuity in the momentum distribution function , in 1d .
it turns out , however , that this distinction , while being profound at a fundamental theoretical level ( because it implies the non - existence of quasiparticles in 1d systems ) , is rather irrelevant to the understanding of the collective mode spectra and their experimental realization in semiconductor quantum wire structures .
as has been emphasized elsewhere , rpa gives a very good account of the collective mode properties in 1d semiconductor quantum wire structures , and we discuss this issue in some details in this paper in the context of the two - component 1d plasma .
the rest of the paper is organized as follows : in sec .
ii we provide the basic rpa formalism as well as the calculated analytic and numerical results for plasmon dispersion , damping , and spectral weight by obtaining the dynamical structure factor ( which is a direct measure of the light scattering intensity ) ; in sec .
iii we study the effects of the interwire electron tunneling in a bi - wire system on the collective charge density excitation spectra ; in sec . iv the plasmon dispersion and damping in the photoexcited 1d ehp is studied in both rpa and including local field corrections ; in sec .
v we briefly discuss the acoustic plasmon mode of the two - component 1d luttinger liquid ; we conclude in sec .
vi providing a brief summary of our results .
in general , the density fluctuation spectra or the plasmons or the longitudinal collective modes of a solid state plasma are given by the poles of the density - density correlation function , or equivalently by the zeros of the dynamical dielectric function .
for a multicomponent plasma the collective modes are given by the zeros of the generalized dielectric tensor @xmath24 .
in the two - component spatially separated 1d plasma the generalized dielectric tensor with wire indices _ i , j _ ( @xmath25 indicating the two quantum wires ) is given ( neglecting any inter - wire tunneling ) within the rpa by @xmath26 where @xmath2 is the 1d wave vector in the @xmath19-direction of free motion , @xmath27 is the mode frequency ( @xmath28 throughout this paper ) , @xmath29 the coulomb interaction in the 1d wire representation , and @xmath30 the 1d non - interacting irreducible polarizability function for the @xmath31-th `` component '' ( i.e. , the @xmath31-th wire with @xmath32 corresponding to the two wires ) . in an ideal 1d electron system the coulomb interaction @xmath33 in the wave vector space is logarithmically divergent , but in our realistic finite width quantum wire model with the confinement width @xmath3 we obtain the intra - wire [ inter - wire ] coulomb interaction matrix element @xmath33 [ @xmath34 by taking the quantizing confinement potential to be of infinite square well type@xcite , @xmath35 @xmath36 where @xmath37 is the background high - frequency lattice dielectric constant , the @xmath38 ( @xmath39 ) sign in eq .
( [ uq ] ) refers to interaction between the same ( opposite ) types of charge in the two wires , ( the two wires could in principle have electrons and holes respectively , rather than being just all electrons or all holes in both wires ) , and @xmath40 , @xmath41 are form factors associated with confinement matrix elements , which are given , for our infinite square well confinement model , by @xmath42 @xmath43 for two wires parallel to each other aligned in the @xmath22 plane . here
@xmath44 is the zeroth - order modified bessel function of the second kind , and the function @xmath45 + 3\sin(2\pi x)/(2\pi)$ ] .
if the two wires are parallel to each other aligned in the @xmath23 plane the inter - wire coulomb form factor becomes [ with @xmath40 remaining the same ] @xmath46 as @xmath47 , @xmath48 , where @xmath49 is a constant ( independent of @xmath15 ) for @xmath40 , but a function of the inter - wire spatial separation @xmath15 for @xmath41 .
note that as @xmath50 @xmath51 becomes @xmath33 , as it should , because the two wires are identical .
the irreducible 1d polarizability @xmath52 in the pure 1d system is given by the bare bubble diagram .
the analytic form of @xmath53 for complex frequency is given by@xcite @xmath54 , \label{pi0}\ ] ] where the principal value of logarithm ( i.e. , @xmath55 < \pi$ ] ) should be taken for complex frequency . in evaluating @xmath52 for real frequency , the usual retarded limit @xmath56 is implied . in eq .
( [ pi0 ] ) @xmath57 and @xmath58 are the effective band mass and the fermi velocity , respectively . in the long wavelength limit ( @xmath1 ) , we can expand the irreducible 1d polarizability as @xmath59 where @xmath60 is the fermi velocity with a fermi momentum @xmath61 for the @xmath62-th 1d component with @xmath63 being the 1d density in the @xmath62-th component . since rpa is known to be valid in the long - wavelength limit ( @xmath64 ) , the limiting forms for the polarizability would be sufficient for our analytic calculations since we are interested in the leading order wave vector dependence of the collective modes of the system .
[ for our numerical results we use the full 1d @xmath65 as given in eq .
( [ pi0 ] ) . ] note that @xmath66 , @xmath67 , @xmath68 , etc .
, refer to the noninteracting or the bare 1d system , and are therefore perfectly well - defined .
the expression for the 1d polarizability @xmath52 given in eq .
( [ pi0 ] ) assumes that the system is pure and free from any impurity scattering . in the presence of impurity scattering characterized by a level broadening or damping @xmath69 ,
the polarizability @xmath70 can be evaluated by means of the standard perturbation theory @xcite to include the impurity scattering induced vertex correction .
the impurity scattering effects are introduced diagrammatically in the rpa by including impurity ladder diagrams into the electron - hole bubble consistently with self - energy corrections in the electron green s function .
since the exact expression for @xmath71 within this diagrammatic approach is complicated , we use in the numerical calculation a particle - conserving approximation for arbitrary values of @xmath2 and @xmath27 , given by mermin @xcite .
the mermin expression is , in fact , equivalent to the diagrammatic result at long wavelength as long as the density of states renormalization by impurity scattering is negligible . in this relaxation time
approximation@xcite , with the impurity scattering induced level broadening @xmath72 , the polarizability is given by @xmath73}. \label{pig}\ ] ] for small @xmath2 and @xmath27 , the polarizability within the mermin formula becomes @xmath74 this approximate formula gives the same long wavelength and low frequency diffusive behavior as one gets from the diagrammatic approach @xcite . in this paper , we take the impurity scattering induced broadening @xmath69 as a constant phenomenological parameter , which , for example , could be taken from the experimentally measured carrier mobility @xmath75 , with @xmath76 .
the condition for the existence of a collective mode is given by the zeros of @xmath77 , the determinant of the dielectric tensor defined by eq .
( [ epsi ] ) . for a two - component system without any tunneling
, we have within rpa : @xmath78 + \left [ v(q)^2 - u(q)^2 \right ] \pi^0_{11}(q,\omega)\pi^0_{22}(q,\omega ) = 0 , \label{zeros}\ ] ] with @xmath79 being the two charge components in confined wires 1 and 2 .
this equation indicates that the sign of the charge components forming the plasma has no effect on the dispersion relation of the collective modes of a two - component system .
the analytical formulae for the long wavelength plasmon dispersion can be obtained even in the general situation where the two charge components have different fermi wave vectors and fermi energies ( by virtue of having , for example , different effective masses @xmath80 and densities @xmath81 ) . in this section
we study the collective modes in this general situation ( i.e. , @xmath82 , @xmath83 , and @xmath84 ) although we should emphasize that such a completely general situation is not physically particularly applicable ( because in most existing quantum wire systems either one has @xmath85 and @xmath82 or @xmath84 and @xmath86 , corresponding respectively to a bi - wire system or an ehp system ) , and in sec .
iv , we investigate , in detail , the modes in a photoexcited homogeneous ( i.e. , @xmath21 ) 1d ehp ( @xmath21 , @xmath87 , @xmath84 ) where the electrons and holes are not spatially separated , but have different effective masses . first , we consider the plasmons of a symmetric two - component system having equal densities ( @xmath88 ) , masses ( @xmath89 ) , and a finite inter - wire separation @xmath15 , and neglecting impurity scattering effects ( @xmath90 ) for simplicity . in this case @xmath91 and eq .
( [ zeros ] ) becomes @xmath92 \times \left[1- 2 v_-(q ) \pi_{11}^0 ( q,\omega ) \right ] = 0 , \label{zeros1}\ ] ] where @xmath93/2 $ ] . by solving eq .
( [ zeros1 ] ) we have two branches of collective modes corresponding to the in - phase op mode @xmath9 and the out - of - phase ap mode @xmath10 ( the in - phase @xmath9 and out - of - phase @xmath10 nomenclature applies to the situation when both components are electrons or both are holes ; otherwise , i.e. , for an electron - hole system , @xmath9 is the out - of - phase mode and @xmath10 the in - phase mode ) : @xmath94 ^ 2 - \left
[ v_f q - e(q ) \right ] ^2}{a_{\pm}(q ) -1 } , \label{wpm}\ ] ] where @xmath95 and @xmath96 $ ] .
the optical mode @xmath9 ( acoustic mode @xmath10 ) indicates the symmetric in - phase ( antisymmetric out - of - phase ) density oscillation of the coupled system assuming the two carriers have the same charge . in the long wavelength limit @xmath1 , eq .
( [ wpm ] ) becomes @xmath97^{1/2},\ ] ] where the asymptotic forms of the coulomb potential @xmath98 as @xmath99 are given by @xmath100 , \nonumber \\
v_-(q \rightarrow 0 ) & = & \frac{2e^2 } { \epsilon_b}\int_0 ^ 1dx \ln \left [ \frac{y(x)}{x } \right ] \left [ ( 1-x)\,(2+\cos(2\pi x ) ) + \frac{3}{2\pi}\sin(2\pi x ) \right ] , \end{aligned}\ ] ] where @xmath101^{1/2}$ ] @xmath102 $ ] for the system aligned in the _ x - z _
( _ x - y _ ) plane . as @xmath47 ,
@xmath103diverge as @xmath104 , but the leading order term in @xmath105 is finite and depends only on the spatial separation @xmath15 . therefore , the leading order term in the 1d op dispersion @xmath106 is enhanced by a factor @xmath107 compared to the one - component system ( corresponding trivially to the change in density from @xmath13 to @xmath108 ) and behaves as @xmath109^{1/2}$ ] as @xmath47 .
the dispersion of the acoustic mode @xmath110 shows a purely linear behavior @xmath111 for @xmath1 with the velocity @xmath112^{1/2 } $ ] as @xmath47 , and lies always above the single particle landau damping region because @xmath113 .
the spatial separation of the two wires produces a weak second order correction to the op dispersion in the two - component system . for @xmath21 ,
the @xmath10 mode becomes degenerate with the electron - hole continuum ( i.e. , @xmath114 ) in the long wavelength limit , and the @xmath9 mode becomes that of the corresponding one - component system with a total density @xmath115 . as the spatial separation increases the optical ( acoustic ) plasmon mode frequency decreases ( increases ) slowly at long wavelengths .
for the system with an infinite separation , the two modes become uncoupled , and have the same dispersion as the 1d plasmon in the one - component system with a total density @xmath13 . in fig .
[ fig0 ] we show our calculated collective modes dispersion for different spatial separations between the two identical quantum wires ; ( a ) @xmath116 and @xmath117 with @xmath118 , and ( b ) @xmath119 and @xmath120 with @xmath121 , where @xmath122 ( @xmath123 ) is the bohr radius of the gaas system and we use this as our length unit throughout this paper . in fig .
[ fig0 ] we use the parameters : @xmath124 and the wire width @xmath125 . as
discussed above , we find that the plasmon dispersion strongly depends on the interwire coulomb correlation determined by @xmath15 . in these figures
we find that for larger spatial separation the energy difference of @xmath126 modes is small because as @xmath15 increases the interwire coulomb correlation decreases .
the important point to note is that both the @xmath126 modes remain undamped upto very large wave vectors due to the suppression of 1d spe , and therefore experimental observation of @xmath126 charge density modes in coupled bi - wire systems should be relatively easy via light scattering or far - infrared spectroscopy .
now we consider the collective mode spectrum in a general system where the two components may have different densities and masses .
the general situation with @xmath21 is of special experimental relevance because it describes the photoexcited 1d ehp system we therefore defer the discussion of the @xmath21 1d ehp system to section iv .
since there is a low energy gap in the 1d spe , we look for undamped plasmons in two regions : the high frequency region ( @xmath127 ) and the low - frequency region ( @xmath128 ) .
( we choose @xmath129 throughout this section without any loss of generality . ) in the long wavelength limit ( @xmath1 ) , we solve eq .
( [ zeros ] ) using eq .
( [ fix ] ) to get @xmath130^{1/2 } , \label{wp}\ ] ] @xmath131^{1/2 } , \label{wm}\ ] ] where @xmath132 . in the long wavelength limit , these asymptotic expressions for @xmath126 are valid in both the high frequency ( @xmath133 ) and the low frequency ( @xmath134 ) regimes .
the condition for the existence of an undamped ap mode ( @xmath10 ) in the high frequency regime(@xmath133 ) is given by @xmath135 , or @xmath136 we find that the condition defined by the inequality [ eq .
( [ cond ] ) ] leads to the conclusion that within rpa @xmath10 can exist as an undamped mode either for a large spatial separation ( and arbitrary @xmath137 ) or for almost equal fermi velocities , i.e. , @xmath138 , ( and arbitrary @xmath15 ) .
as @xmath137 decreases the system should have a large spatial separation to have an undamped high energy ap mode .
for @xmath139 ( e.g. , equal mass and density ) , we find that the acoustic mode lies in the high frequency regime and is always undamped for arbitrary @xmath15 .
this is also true in 2d bilayer systems with finite layer separations , where the ap mode is undamped as long as @xmath139 and @xmath82 .
the condition for the existence of the ap mode in the low frequency regime ( @xmath140 ) is given by @xmath141 .
due to the energy gap in the 1d spe the ap mode is undamped in this regime .
the existence of the undamped ap mode in the low frequency regime for arbitrary @xmath137 and @xmath15 is a unique phenomenon in the 1d two component system arising entirely from the non - existence of low energy spe in 1d . in 2d or 3d two - component system
the low frequency regime is necessarily accompanied by landau damping due to the presence of the low energy spe continua . in fig .
[ fig1 ] we show our numerically calculated longitudinal collective modes ( for different @xmath15 and @xmath137 ) with a fixed density @xmath142 , effective mass @xmath143 ( corresponding to the effective electron mass of the gaas ) , and a wire confinement @xmath144 . in these figures
the higher ( lower ) energy mode @xmath9 ( @xmath10 ) denotes the optical ( acoustic ) mode . in fig .
[ fig1](a ) we use the parameters : the spatial separation of the two wires @xmath145 and @xmath146 .
( note that @xmath137 determines the density of the component two for a given @xmath147 and @xmath148 , @xmath149 .
for example , if @xmath150 we have @xmath151 . ) with these parameters the velocity of the acoustic mode @xmath152 exceeds the fermi velocity of the component one ( @xmath153 ) and lies well above the spes of both components ( high frequency regime ) . in fig .
[ fig1](b ) we choose the parameters which limit the @xmath152 between @xmath154 and @xmath155 : @xmath116 and @xmath156 .
( note that this requirement restricts @xmath15 to unrealistically small values . )
thus , the undamped ap mode with these conditions lies in low frequency regime ( @xmath157 ) which corresponds to the regime between two spes .
the important point is that in both cases ( `` large '' @xmath15 and @xmath137 , and `` small '' @xmath15 and @xmath137 ) the ap is undamped in 1d in contrast to higher dimensional systems .
next , we investigate the plasmon modes of the two - component system in the presence of impurity scattering . we calculate the plasmon dispersion of a spatially separated 1d two - component system with equal masses and densities ( @xmath139 ) in the presence of impurity scattering ( characterized by a constant level broadening @xmath69 ) . with the approximation for the polarizability function for small @xmath2 and @xmath27 defined by eq .
( [ piga ] ) the plasmon dispersion is given by @xmath158 where @xmath9 ( @xmath10 ) indicates the op ( ap ) mode .
note that for @xmath159 , eq . ( [ wgam ] ) recovers the result previously obtained for the pure two - component system .
thus , in the presence of impurity scattering the plasmon modes acquire a wave vector dependent imaginary part which corresponds to a plasma damping , @xmath160 .
the plasmon becomes overdamped at small @xmath2 as it becomes completely imaginary for small @xmath2 , i.e. , the plasmon modes are totally overdamped below the critical wave vectors @xmath161 and exist only for wave vectors @xmath2 larger than these critical values .
( note that the plasmon broadening @xmath162 is , in general , determined by the imaginary part of the complex plasmon mode , and is different from the level broadening @xmath69 . ) for @xmath163 the plasma dampings are roughly comparable to the half of the impurity scattering , i.e. , @xmath164 . from eq .
( [ wgam ] ) we find the critical wave vectors ( below which the plasmon is no longer a well defined collective mode ) as @xmath165 where @xmath166 .
as the impurity scattering rate @xmath69 increases the critical wave vectors @xmath161 also increase .
the @xmath167 ( just as @xmath9 itself ) depends weakly on @xmath15 , but @xmath168 ( just as @xmath10 ) is strongly affected by @xmath15 . as @xmath50
the intra- and inter - wire coulomb interactions become the same , @xmath169 , and consequently @xmath170 . thus , for the system with a large impurity scattering and a small spatial separation we have only the op mode in the long wave length limit . in general , we have @xmath171 from eq .
( [ qcpm ] ) , which means that the ap mode is easily affected by impurity scattering ( and is therefore more difficult to observe experimentally . ) in fig .
[ fig2 ] we show the plasmon dispersion calculated numerically by using the full polarizability , eq .
( [ pig ] ) , in the presence of finite impurity scattering .
we find zeros of the complex dielectric function of the 1d two - component system in the presence of impurity scattering , i.e. , @xmath172 , to obtain the plasmon dispersion curves . in fig .
[ fig2 ] the curves with @xmath173 give the plasma frequency or the real part [ @xmath174 , and those with @xmath175 give the plasma damping or the imaginary part ( @xmath137 ) of the complex zero solutions [ @xmath176 .
the figure shows that the plasmons are overdamped below critical wave vectors @xmath177 .
the critical wave vectors , @xmath161 , below which the plasmon does not exist due to impurity scattering effects , depend on the density and the impurity scattering rate .
we note that the overdamping of the plasmon mode occurring for small @xmath2 is a direct consequence of the diffusive nature of the electron dynamics .
we emphasize that this long wavelength impurity scattering - induced suppression of collective modes is more severe in 1d than in 2d or 3d ( in 3d , because the op is gapped , it is hardly affected by impurity scattering . ) finally , we study the dynamical structure factor , @xmath178 $ ] , which gives a direct measure of the oscillator strength of the density fluctuation spectrum , and many experimental probes such as inelastic electron and raman scattering spectroscopies are directly related to the dynamical structure factor @xcite .
equivalently , the dynamical structure factor is a measure of the spectral weight carried by the particular collective mode of the electron liquid .
an undamped plasmon shows up as a @xmath179-function peak in @xmath180 indicating the existence of a simple zero of @xmath181 . when both re@xmath182 $ ] and im@xmath182 $ ] become zero [ i.e. , @xmath172 ] the imaginary part of the inverse dielectric function im@xmath184 $ ] is a @xmath179-function with the strength @xmath185 / \partial \omega|_{\omega=\omega_{\pm}(q)}},\ ] ] where @xmath186 are the plasmon dispersions for the system .
the damped plasmon ( @xmath187 ) , however , corresponds to a broadened peak in @xmath180 for larger broadening , as for @xmath188 in the system with impurity scattering , the plasmon is overdamped and there is no peak in @xmath180 . at long wavelengths ( @xmath1 )
, the optical plasmon @xmath9 mostly exhausts the @xmath189-sum rule and carries almost all the spectral weight .
@xmath190 approaches zero as @xmath47 because the 1d plasma frequency vanishes .
note that the small weight of the ap mode at long wavelengths makes it particularly susceptible to collisional damping effects in 1d two component system .
for the system with the same effective mass and a large spatial separation , the acoustic plasmon mode has an energy comparable to that of the optical plasmon mode , which gives the enhanced spectral weight associated with the ap mode . in fig .
[ fig3 ] the im@xmath184 $ ] is plotted for different wave vectors as a function of the frequency for two different systems . in fig .
[ fig3](a ) we choose the parameters such that the ap mode lies in the low frequency regime : @xmath191 , @xmath192 , @xmath193 , and @xmath194 . in fig .
[ fig3](b ) we choose the parameters such that the ap mode lies in the high frequency regime : @xmath195 , @xmath196 , @xmath197 , and @xmath198 .
the insets in fig .
[ fig3 ] show the weight @xmath190 of the plasmon modes . for pure system the plasmon peaks become @xmath179-function peaks with weights given in the inset , but with impurity scattering rate @xmath199 and @xmath200 , we show broadened peaks giving the damped plasmon modes . as shown in fig .
[ fig3 ] the spectral weight of the acoustic mode is comparable with that of the optical mode at finite wave vectors , but not in the long wavelength limit . with the enhanced spectral weight
the acoustic plasmon mode in 1d should be more easily observable than in higher dimensions and may significantly affect physical properties of the quasi-1d two - component systems .
the results presented in this section indicate that the ap mode should be observable in double wire structures under experimentally realizable conditions .
in this section we analytically calculate the collective mode dispersions in bi - wire structures ( @xmath201 , i.e. , @xmath202 and @xmath203 , and @xmath82 ) _ in the presence of significant interwire quantum tunneling _ , all earlier work in the previous section having dealt with the zero tunneling limit . since
our focus is on understanding tunneling effects in this section , we put @xmath204 with no loss of generality . in the presence of interwire tunneling , the electron energy eigenstates @xmath205 should be used @xcite as the basis set rather than the wire index which is no longer a good quantum number .
the energy levels @xmath206 , where @xmath207 is the parabolic one electron 1d kinetic energy in each wire and @xmath14 is the tunneling strength , are the usual symmetric and antisymmetric one electron eigenstates in the presence of tunneling with the single particle symmetric - antisymmetric ( sas ) gap given by @xmath208 . in the sas representation
the collective mode spectra become decoupled by virtue of the symmetric nature of our bi - wire system ( i.e. , both wires identical with equal electron density ) , and the collective density fluctuation spectra are given by the following two equations for the in - phase and the out - phase plasmon modes @xmath126 respectively : @xmath209 = 0 , \label{ep}\ ] ] and @xmath210 = 0 , \label{em}\ ] ] where @xmath211 with @xmath33 and @xmath51 being respectively the intrawire and interwire coulomb interaction matrix elements given in eqs .
( [ vq ] ) and ( [ uq ] ) .
the @xmath212 in eqs .
( [ ep ] ) and ( [ em ] ) , with @xmath213 , are the noninteracting sas polarizability functions within our rpa theory : @xmath214 where @xmath215 are fermi occupancy factors . solving eqs .
( [ ep])([piab ] ) we obtain the collective density fluctuation spectra of the coupled bi - wire system . in the absence of tunneling , @xmath216 , one has @xmath217 , and one then recovers in a straightforward fashion the results we have in sec .
ii , i.e. , the in - phase optical ( @xmath218 ) and the out - of - phase acoustic ( @xmath11 ) plasmons of a bi - wire system without any electron tunneling .
it is , in fact , straightforward to obtain analytically [ from eqs .
( [ ep])([piab ] ) ] the long wavelength ( @xmath1 ) plasma modes of the coupled bi - wire system _ including effects of interwire tunneling_. we obtain in the long wavelength limit the following results : @xmath219 , \label{wpi}\ ] ] @xmath220 where @xmath221 is the fermi velocity and @xmath222 for @xmath223 when both symmetric and antisymmetric levels are occupied ( i.e. , the 1d fermi energy @xmath224 ) , and @xmath225 for @xmath226 when only the symmetric level is occupied ( i.e. , @xmath227 ) .
the most important qualitative feature of the plasmon dispersion in the presence of interwire tunneling is that even though the in - phase op mode ( @xmath9 ) depends weakly on tunneling , the out - of - phase mode ( @xmath10 ) , which is purely acoustic in the absence of tunneling ( @xmath11 if @xmath228 ) , develops a plasmon gap at @xmath12 in the presence of nonzero tunneling .
the plasmon gap @xmath229 depends nontrivially on the 1d electron density @xmath13 , the interwire tunneling amplitude @xmath14 , and the interwire distance @xmath15 .
it is easy to see from eq .
( [ wmi ] ) that this plasmon gap @xmath230 , has the following behavior @xmath231 depending on whether the interwire tunneling is strong or weak .
it should be emphasized that the strikingly non - intuitive @xmath232 dependence of the collective mode gap ( on the square root of the single particle gap ) is purely a coulomb interaction effect , which dominates the collective excitation spectra in the weak tunneling situation .
we mention the curious phenomenon that interwire tunneling converts the @xmath10 mode into an effective optical mode by opening up a long wavelength plasmon gap ( for @xmath216 the @xmath10 mode is the ap mode ) .
in sections ii and iii we discussed a double wire system where the two one dimensional charged components are separated by a distance @xmath15 .
now we consider the possibility of acoustic plasmons in a photoexcited homogeneous ( i.e. , @xmath21 ) 1d ehp where the electrons and holes are not spatially separated .
this is , in fact , the original context@xcite in which the ap mode was first discussed in the solid state plasma more than forty years ago .
the photoexcited ehp in semiconductors is also the system in which most of the early investigations @xcite of acoustic plasmons were carried out .
our reason for singling out the 1d ehp for a detailed investigation in this section ( in principle , it is just the @xmath21 limit of sec .
ii ) is the existence of 1d ehp in real samples @xcite from several groups where a search for 1d @xmath126 modes via light scattering spectroscopy should be successful .
it has been known for a long time @xcite that a two - component ehp allows for the existence of two collective modes , one of which ( the low frequency one ) is linear in wave vector or acoustic at long wavelengths , and the other mode ( the high frequency or the optical plasmon ) is essentially the combined collective charge density oscillation of the whole system being therefore qualitatively similar to the usual plasmon mode in a single component plasma . in an electron - hole two - component plasma , as appropriate in a photoexcited semiconductor system
, the ap corresponds to the in - phase collective density fluctuation excitation of electrons and holes , and the op corresponds to their out - of - phase motion with respect to each other .
[ if the two components are both electrons or both holes , for example , in a bi - wire ( as considered in sec .
ii of the paper ) or bilayer or a two - band system , then the ap signifies the out - of - phase collective mode and the op the in - phase motion in general , the acoustic mode is the neutral mode at long wavelengths and the optical mode the charged mode . ] in spite of compelling theoretical arguments @xcite for the existence of `` quantum '' ap in degenerate two - component homogeneous electron - hole solid state plasmas the experimental observation of low temperature ap in two and three dimensional homogeneous electron - hole semiconductor plasmas has been very severely hindered by landau damping effects .
it turns out that in two and three dimensional degenerate homogeneous ehp the ap is necessarily landau damped ( by decaying into single particle excitations ) and is in fact unobservable as a well - defined mode unless the effective mass ratio between the two species is extremely large @xcite .
this has prevented an unambiguous observation @xcite of a quantum ap in degenerate solid state ehp although the corresponding classical ap ( in the non - degenerate high temperature electron - ion plasma ) was experimentally observed @xcite almost forty years ago . in this section
we show that the elusive quantum ap in the homogeneous ( i.e. , @xmath21 ) ehp should be relatively easy to observe in the degenerate 1d ehp @xcite confined in semiconductor quantum wire structures due to the severe suppression of single particle excitations in one dimensional systems
. in particular , within the rpa which should be quantitatively valid in one dimension by virtue of the smallness of vertex corrections , the ap mode in a 1d ehp in gaas quantum wires should be _ completely _ undamped upto a rather large wave vector @xmath233 , in contrast to two / three dimensional ehp where the ap is damped in gaas and is essentially unobservable .
we believe that resonant inelastic light scattering spectroscopy , which has already been successfully used @xcite in observing the 1d plasmon in _ single component _ ( @xmath13-doped ) gaas quantum wires , is the ideal tool in the search for the ap in _ photoexcited _ ( undoped ) two component ehp in gaas quantum wires .
theoretical results presented in this section show that the 1d in - phase ap mode should be readily experimentally observable in resonant inelastic light scattering experiments performed on photoexcited 1d ehp @xcite in narrow gaas - algaas quantum wires .
we put @xmath21 in the formalism developed in sec .
iii of the paper and allow for the possibility of different masses @xmath234 and @xmath235 taking equal densities @xmath236 of the electrons and holes .
below we present and discuss our results for collective modes in 1d ehp both in the rpa and in the hubbard approximation ( ha ) where one includes a local field correction to the rpa non - interacting polarizability .
the inclusion of local field corrections ( ha ) beyond rpa distinguishes this section from sec .
ii where strict rpa was used our motivation for using ha ( as well as rpa ) is to make our results more realistic for the experimental systems where local field corrections may very well be important at the experimentally feasible densities and wave vectors .
the dynamical dielectric function in the ha is given by @xmath237\pi_e^0(q,\omega ) \right \ } \left \ { 1 - v(q)[1-g_h(q)]\pi_h^0(q,\omega ) \right \ } \nonumber \\ & & - v(q)^2\pi_e^0(q,\omega)\pi_h^0(q,\omega ) , \label{pih}\end{aligned}\ ] ] where @xmath238 is a simple static local field correction to the rpa and is given by @xmath239 , where @xmath240 are the fermi wave vector of the electron and hole respectively .
the calculation in the ha is similar to that in the rpa with the rpa dielectric function [ eq . ( [ epsi ] ) ] being replaced by the ha dielectric function [ eq . (
[ pih ] ) ] . by putting @xmath21 we have @xmath241 , @xmath242 .
thus , in the long wavelength limit @xmath47 , we have the plasmon mode dispersions within rpa from eqs , ( [ wp ] ) and ( [ wm ] ) @xmath243^{1/2 } , \label{wph}\ ] ] @xmath244 where @xmath245 corresponds to the mass ratio @xmath246 since @xmath247 . as the mass ratio decreases ( @xmath248 ) , the op mode ( @xmath9 ) becomes the plasmon mode of the one component system .
since @xmath249 we have @xmath250 .
the ap mode within rpa lies always in the undamped region ( low frequency regime ) between two spes .
this undamped 1d ap mode @xmath10 in the long wavelength limit is fundamentally different from the ap mode of the higher dimensional ehp since the presence of the low energy landau damping region in 2d or 3d gives rise to the complete damping of ap mode , unless @xmath137 is very small in the 1d ehp the ap mode is undamped within rpa for arbitrary values of @xmath137 ( @xmath251 by choice ) . in figs .
[ fig4 ] [fig6 ] we show the numerically calculated plasmon modes and spectral weights for different densities in the photoexcited gaas quantum wire 1d ehp within both rpa and ha . in figs .
[ fig4 ] [fig6 ] we use the parameters corresponding to gaas : @xmath252 , @xmath253 , @xmath254 , and the confinement width of the wire @xmath125 .
we also use a level broadening @xmath255 due to the impurity scattering in calculating the mode spectral weights . note that in the long wavelength limit the rpa ap modes are undamped ( except , of course , for impurity damping ) for all densities because of the absence of long wavelength landau damping . in general ,
similar to the corresponding higher dimensional situations the local field effects in the ha reduce the plasma frequencies @xcite compared with the rpa results , in particular , in the low density system . since the @xmath9 mode exhausts the @xmath189-sum rule in the long wavelength limit , the local field correction has relatively little effect on the @xmath9 mode .
the @xmath10 ap mode is , however , very strongly affected by local field effects .
[ fig4 ] shows the results for the ehp density @xmath256 : at this density we find the remarkable result that the @xmath10 ap mode is completely suppressed by local field effects as it gets pushed into the spe landau damping continua .
thus , in fig .
[ fig4](a ) the ap mode dispersion within ha does not show up because @xmath10 lies completely in the landau damping continuum .
[ fig4](b ) shows that within rpa ( solid lines ) the @xmath10 mode ( low energy peaks ) carries an appreciable spectral weight which is comparable to that carried by the op mode @xmath9 ( high energy peaks ) , however , within ha ( dashed lines ) the @xmath9 mode actually carries all the spectral weight ( we find only the high energy @xmath9 peaks in the ha with the low energy @xmath10 peaks being completely suppressed . ) as density increases the ( completely suppressed ) @xmath10 mode in ha reappears in the low energy regime as the ap mode comes out of the landau continuum . in figs .
[ fig5 ] and [ fig6 ] we show the results for density @xmath257 and @xmath258 , respectively . at the density @xmath259
the @xmath10 mode reappears below the spe@xmath260 in the ha .
this @xmath10 mode lying below spe@xmath260 is undamped due to the low energy spe gap in 1d and is a characteristic feature of 1d ehp .
the spectral weight of the undamped @xmath10 mode in the low energy region ( at higher density ) is comparable to that of the @xmath9 mode even within ha [ fig . [ fig5](b ) ] .
at very high density ( fig .
[ fig6 ] ) @xmath258 the local field corrections are weak , and we find that the @xmath10 mode ( in the ha ) lies between spe@xmath261 and spe@xmath260 and is undamped in the long wave length limit .
the calculated spectral weight in the ha is comparable to that in the rpa in these high density systems .
one conclusion of our ha - based calculations is that one should use higher ehp densities for an unambiguous observation of undamped 1d ap this is consistent with our rpa results also , where landau damping is absent , because in rpa higher carrier densities imply quantitatively weaker effects of impurity level broadening .
given the unique nature of the spe continua ( and the strong low energy suppression of landau damping ) in the 1d system , the question naturally arises about whether the collective mode behavior in 1d ehp is fundamentally different from that in 2d and 3d systems , where the ap mode is usually unobservable due to its unavoidable decay via landau damping to the electron(i.e . , the lighter mass component )
spe unless @xmath137 is very small ( i.e. , the mass ratio very small , @xmath262 ) which cause the landau damping to be small , but still non - zero . in the 1d ehp ,
however , there are three ( as opposed to only one in 2d and 3d ehp ) landau damping - free regimes in the @xmath27-@xmath2 space these are [ see fig .
5(a ) or 6(a ) or 7(a ) ] the high energy - low wave vector regime above the lighter mass spe continuum ( which is present in 2d and 3d systems also ) , the low energy regime below both the spe continua , and the intermediate energy low to intermediate wave vector regime in the spe gap between the spe continua of the two components ( we emphasize that landau damping of collective modes is allowed in 2d and 3d systems in the last two regimes even in the rpa because there is no low energy gap in the spe continuum in higher dimensions ) . to emphasize the qualitative difference between 1d ehp and its higher dimensional counterparts with respect to the landau damping characteristics of the collective modes , we provide in figs . 8 and 9 our calculated rpa results for collective mode dispersion and damping in 2d and 3d ehp respectively , choosing the same effective mass , lattice dielectric constant , level broadening , and `` approximately equivalent '' carrier densities .
one can see that in figs .
8(a ) , the 2d ehp case , and 9(a ) , the 3d ehp case , the op mode is above the spe continua of both components ( exactly the same as in figs . 5 7 for the op in 1d ehp ) and is therefore not landau damped whereas the ap mode in both case lies within the electron ( i.e. , the lighter mass ) spe continuum ( but above the hole spe ) and is therefore landau damped by decaying into the quasiparticles of the electron component in the ehp .
the contrast between the rpa results for the 2d/3d ap mode ( figs . 8 and 9 ) and the 1d ap mode ( figs . 5 - 7 ) is striking in terms of landau damping the 2d/3d ap mode is invariably landau damped by being inside the spe continuum of the lighter carrier component ( i.e. , the higher fermi velocity , @xmath263 , in our case ) where the 1d ap mode ( within rpa ) is undamped by being in the intermediate regime in between ( the spe `` gap '' regime ) the spe continua of the two components and is therefore _ not _ landau damped by either plasma component .
this non - existence of ap landau damping in the spe `` gap '' regime in a 1d ehp is a fundamental one dimensional effect arising from the suppression of long wavelength single particle excitations peculiar to 1d .
we should mention , however , that , as is obvious from fig .
5 - 7 , inclusion of local field corrections within the ha may push the ap mode down into the spe continua of the higher mass plasma component , i.e. , the hole spe , making the ap mode strongly landau damped even in the 1d ehp our comparison between 1d and 2d/3d ehp collective modes and their qualitative difference are based on the rpa , going beyond rpa may very well render the 1d ap unobservably landau damped as happens , for example , within the ha in our calculation for @xmath264 @xmath265 in figs . 5 and 6 .
it is very anticlimactic , however , that even within the rpa the qualitative difference in the landau damping behavior of the ap in the 1d ehp compared with its higher dimensional counterparts is not dramatically reflected in the mode spectral weight or the dynamical structure factor , as can be seen by comparing the rpa spectral weights in fig .
5(b ) , 6(b ) , 7(b ) in the 1d ehp with those in the 2d/3d ehp given in figs .
8(b)/9(b ) .
( we emphasize that we use equivalent impurity level broadening , @xmath266 , in all our spectral weight calculations . ) in general , the spectral weight carries by the ap mode ( the lower frequency peak in the dynamical structure factor im[@xmath267 for each @xmath2 ) is comparable , albeit slightly larger , in the 1d ehp as that in the 2d/3d ehp . in particular , the ap spectral weights in the 1d and the 2d ehp seem to be quantitatively similar whereas the corresponding 3d ap mode carries considerably less spectral weight ( which is understandable based on the fact that in 3d the op has a finite gap at long wavelengths in contrast with the 1d/2d op which vanishes at long wavelengths , and therefore in a 3d ehp the long wavelength op mode carries considerably more relative spectral weight than its lower dimensional counterparts ) .
the fact that the 1d ap carries small spectral weight in spite of its lack of landau damping can be easily understood on the basis of the @xmath189-sum rule , which is essentially completely exhausted by the op in the long wavelength limit , leaving rather little spectral weight to be carried by the 1d ap in the long wavelength . at large wave vectors ( i.e. , away from the long wavelength limit ) , on the other hand
, the ap is landau damped ( because it enters the spe continua ) and all collective modes start carrying little spectral weights .
we therefore conclude this section on 1d ehp collective modes with two relatively modest conclusions : ( 1 ) in spite of the non - existence of ap landau damping in the 1d ehp within rpa , the calculated ap spectral weight is not much larger than the corresponding 2d situation ( it is , however , substantially larger than the corresponding 3d situation ) , and therefore the observability of 1d ehp ap should be only slightly more favorable than the corresponding 2d case ; ( 2 ) inclusion of local field corrections in general severely suppresses the ap spectral weight in the 1d ehp , and at least within the ha the ap becomes unobservable at lower densities .
in this section we briefly discuss , for the sake of completeness , the ap mode of the two - component luttinger liquid , which has two different fermi velocities @xmath263 and @xmath268 corresponding to the electron and the hole respectively .
we make the standard luttinger model approximation of linearizing the single particle energy .
this system can be described by the 2-component luttinger model @xcite hamiltonian @xmath269 + \frac{1}{2l } \sum_{i , j , q } v_{i , j}(q ) \rho_{i}(q)\rho_j(-q ) , \label{hl}\ ] ] where _ i , j_=_e , h _ , and @xmath270 ( @xmath271 ) is the destruction operator of the right- ( left- ) moving @xmath62-th component with momentum @xmath272 and spin @xmath273 , @xmath274 the coulomb interaction between @xmath62-th and @xmath31-th component ( here @xmath275 , and @xmath276 the total density operator of the @xmath62-th component . by standard bosonization methods @xcite we can diagonalize eq .
( [ hl ] ) and find two eigenmodes corresponding to the charge density excitations @xmath277 \nonumber \\ & \pm & \frac{q^2v_{fe}^2}{2 } \left \ { \left [ 1 + \alpha^2 + ( 1+\alpha)\tilde{v}(q ) \right ] ^2 - 4\alpha \left [ \alpha + ( 1+\alpha)\tilde{v}(q ) \right ] \right \}^{1/2 } , \label{wpml}\end{aligned}\ ] ] where @xmath278 and @xmath279 ; note that @xmath280 in the 1d ehp . in the long wavelength limit , @xmath1 , eq . ( [ wpml ] ) becomes @xmath281^{1/2 } , \label{wpl}\ ] ] @xmath282 these long wavelength charge density modes defined by eqs . ( [ wpl ] ) and ( [ wml ] ) are exactly the same as the long wavelength collective modes in the 1d ehp within rpa [ eqs .
( [ wph ] ) and ( [ wmh ] ) ] .
this can be explained by the fact that vertex corrections to the irreducible polarizability of the 1d system vanish @xcite in the luttinger model , and therefore rpa and luttinger model results become identical in the long wavelength limit . even though the charge density mode dispersions of the luttinger model are identical to the rpa collective mode predictions based on the fermi liquid theory , the damping properties of the @xmath10 ap mode in the luttinger model is strikingly different form the rpa result . in the previous section we found that within rpa the ap mode in the 1d ehp exists only at long wavelengths as the ap mode
gets severely damped by landau damping at finite wave vectors where it enters the electron single particle continua and is landau damped . however , since there are no equivalent spe landau damping mechanisms in the luttinger model the ap mode in the luttinger model seems not to be damped out even at the finite wave vectors , and seems to exist at arbitrary wave vectors .
the above - discussed difference between the rpa ( fermi liquid ) result and the luttinger model ( non - fermi liquid ) result seems to suggest a direct experimental approach to observe the predicted non - fermi liquid - like behavior of a one dimensional system by studying its plasmon damping behavior which , according to the rpa ( luttinger ) model , should ( should not ) manifest finite wave vector landau damping .
this is , however , misleading because the strict non - existence of any damping of the collective modes at any wave vectors is purely a result of the luttinger _ model _ ( rather than being a generic luttinger _ liquid _
result ) , arising entirely from the linearization of the single particle energy [ the first term in eq .
( [ hl ] ) ] in the hamiltonian . in the generic luttinger liquid behavior , where one does _ not _ linearize the single particle energy , higher order processes ( which asymptotically vanish at low energies and long wavelengths ) produce collective mode damping arising from the ( multiboson ) boson - boson collision processes within the bosonized theoretical description . such multiboson collision processes , which are akin to phonon - phonon anharmonic scattering processes producing phonon decay beyond the harmonic crystal theory , are _ irrelevant _ in the renormalization group sense at long wavelengths , but do produce damping of the collective modes at finite wave vectors . in this sense , therefore , no fundamental distinction exists between the two theoretical approaches since both predict damping of the collective modes at finite wave vectors , but _ not _ at zero wave vectors .
one expects the ap damping in the luttinger liquid theory to be a smooth function of wave vector @xmath2 , vanishing only in the long wavelength limit @xmath1 .
this seems to be drastically different from our rpa damping result where the ap is undamped upto a wave vector @xmath283 when it becomes damped as it enters the spe continuum .
going beyond rpa , however , one finds that the ap is damped even outside the spe landau damping continua due to multipair ( i.e. , multiple electron - hole excitations ) productions , and therefore there may _ not _ be much qualitative difference between rpa and luttinger liquid theories even for the ap damping properties .
this issue , however , merits further investigations , and detailed theoretical calculations of the ap damping properties in the two - component luttinger liquid may very well point to some significant differences with our rpa results , which then may lead to a direct method of differentiating between luttinger liquid and rpa results , based on the experimentally measured damping properties . very recently , such a calculation @xcite has been carried out for a _
one component _
clean luttinger liquid using the _ short range _ interaction which , however , has to be generalized to the realistic long - range coulomb interaction in 1d quantum wires before any firm conclusion on the damping behavior contrast between rpa and luttinger liquid may be reached .
we emphasize that the ap damping property being manifestly a non - universal property ( determined by irrelevant operators such as the band curvature ) , its actual quantitative calculation in the luttinger liquid theory is problematic .
we have obtained the dispersion and spectral weight of collective charge density excitations in zero temperature two - component one dimensional quantum plasma confined in semiconductor quantum wire structures . in some sense
this paper is a two - component generalization of our earlier work @xcite where we studied the zero temperature collective charge density excitations of the one component 1d system in some details . in a separate earlier publication @xcite
we consider a specific example of collective modes in a two - component 1d electron system ( namely , a two - subband 1d system ) in the context of obtaining quantitative agreement with the experimental results of ref . .
in the current paper we refrain from repeating any results from these earlier publications @xcite of ours , and the current paper , along with refs . and , forms a reasonably complete quantitative description of possible low lying collective charge density excitations in 1d semiconductor quantum wire structures .
our main results in this paper are the following : ( 1 ) there are two possible collective charge density excitations in a two - component 1d quantum plasma associated with the relative phase fluctuations in the two charge densities the op mode is the in - phase ( out - of - phase ) mode when the two components are the same ( opposite ) kind ( both electrons or both holes versus one component electron and the other component hole ) and the ap is the out - of - phase ( in - phase ) mode in the same situation ( in general , the ap is essentially a neutral excitation at long wavelengths and the op is the collective oscillation of the total charge density ) ; ( 2 ) in contrast to higher dimensional systems where the ap mode is often landau damped because it lies in the spe continuum of the faster moving charge component , the 1d ap mode is invariably undamped at long wavelengths ( within rpa ) due to the severe suppression of long wavelength spe continua in 1d ; ( 3 ) this rpa result of the non - existence of long wavelength landau damping in the 1d ap mode is drastically affected by local field corrections , which at low density may overdamp the ap mode by pushing it inside the spe continua of both components the op mode on the other hand is relatively robust with respect to local field corrections since its long wavelength dispersion is fixed by the @xmath189-sum rule ; ( 4 ) the ap mode in general carries little spectral weight at long wavelengths , and even at finite wave vectors its spectral weight is only slightly enhanced with respect to the corresponding 2d two - component situation , leading to the somewhat disappointing and unexpected conclusion that in spite of the complete suppression of the long wavelength spe continua in 1d systems the neutral ap mode is not substantially easier to observe in 1d than in 2d/3d systems ; ( 5 ) finite inter - wire separation in general enhances the ap spectral weight increasing its observability ; ( 6 ) finite impurity scattering induced level broadening gives rise to a critical wave vector below which the collective modes are always overdamped , and thus the concept of a long wavelength collective mode does not strictly apply ; ( 7 ) finite inter - wire tunneling produces a long wavelength plasma gap in the ap mode ; ( 8) luttinger liquid theory and fermi liquid - based rpa theory produce formally identical op and ap collective mode behavior in 1d two component systems making it impossible to observe any characteristic 1d luttinger liquid signature in the collective charge density excitation spectral in one dimension ; ( 9 ) the ap mode being strongly ( and qualitatively ) affected by the local field corrections ( with the op mode being relatively unaffected ) , in principle , it should be possible to investigate many body effects in 1d systems by studying the dispersion and damping of the ap mode ( the op mode on the other hand should be relatively insensitive to these many body corrections except at rather large wave vectors ) . in our calculations
we have concentrated on two distinct types of experimentally realizable two - component quasi-1d quantum plasmas confined in semiconductor ( gaas ) quantum wire nanostructures : spatially separated doped quantum double - wire structures and photoexcited homogeneous 1d ehp in a single wire .
both systems are potentially interesting from the perspective of 1d collective excitation spectra and provide somewhat complementary information .
our hope is that our detailed quantitative ( both numerical and analytical ) investigation of the collective mode dispersion and spectral weight in two - component 1d system will motivate experimental work ( involving inelastic light scattering and far infrared frequency domain spectroscopies ) on the subject searching for the 1d ap mode in quantum wire structures .
this work is supported by u.s .- onr and the u.s .- aro . for reviews
see , e.g. _ nanostructures and mesoscopic systems _ , edited by w. p. kirk and m. a. reed ( academic , new york , 1992 ) ; c. w. j. beenakker and h. van houten , in _ solid state physics : advances in research and applications _ , edited by h. ehrenreich and d. turnbull ( academic , new york , 1991 ) .
a. r. goi , a. pinczuk , j. s. weiner , j. m. calleja , b. s. dennis , l. n. pfeiffer , and k. w. west , phys .
rev . lett . *
67 * 3298 ( 1991 ) ; a. r. goi , _ et al .
_ , in _ phonons in semiconductor nanostructures _ ,
, edited by j. p. leburton , j. pascual , and c. s. torres ( plenum , new york , 1993 ) , p. 287 .
p. m. platzmann and p. a. wolff , _ waves and interactions in solid state plasma _
( academic , new york , 1973 ) , chap . 5 ; m. v. klein , in _ light scattering in solids i _ ,
edited by m. v. cardona ( springer , berlin , 1975 ) , p. 147 ; j. ruvalds , adv .
phys . * 30 * , 677 ( 1981 ) .
|
collective charge density excitation spectra of both a spatially separated two - component quasi - one dimensional ( 1d ) quantum plasma , as existing in a semiconductor double quantum wire structure , and a 1d homogeneous electron - hole plasma , as appropriate for a photoexcited semiconductor quantum wire system , are calculated within the two - component random - phase - approximation .
we find two phase fluctuation collective modes , one ( optical plasmon , op ) with energy proportional to @xmath0 is the _ total _ in - phase ( out - of - phase ) charge density oscillation of the system and the other ( acoustic plasmon , ap ) with a linear energy dispersion as @xmath1 is the _ neutral _ out - of - phase ( in - phase ) charge density oscillation of the system for the situation where the two components have the same ( opposite ) charges , where @xmath2 is the 1d wave vector and @xmath3 is a characteristic 1d confinement size .
in contrast to higher dimensional systems we find the neutral long wavelength ap mode to be generically undamped by landau damping effects due to the severe suppression of single particle excitations in 1d systems .
we also investigate the effect of impurity scattering on the collective mode dispersion and damping , and calculate the collective mode spectral weight by obtaining the dynamical structure factor .
we find that both op and ap modes are overdamped by impurity scattering below some critical wave vector .
we find that in the long wavelength limit the spectral weight is carried mostly by the op , but the spectral weight of the undamped ap mode at finite ( but not too large ) wave vectors is comparable with that of the op mode , making it viable to observe the ap mode in semiconductor quantum wire systems .
the effect of the interwire electron tunneling in a biwire system on the collective charge density excitation spectra is also studied .
we discuss the mode dispersion and damping from an effective luttinger liquid perspective as well and in some cases include local field corrections in our calculations .
pacs number : 73.20.mf ; 71.35.ee ; 71.45.gm ; 71.45.-d
|
in the standard model ( sm ) , only one higgs doublet is introduced and one neutral higgs boson remains after electroweak symmetry breaking . in the minimal supersymmetric standard model ( mssm )
@xcite , two higgs doublets are required to break the electroweak symmetry .
the two higgs doublets , @xmath2 and @xmath3 , couple to fermions with weak isospin @xmath4 and @xmath5 respectively @xcite . after spontaneous symmetry breaking , there remain five physical higgs bosons : a singly charged higgs boson @xmath6 , two neutral cp - even scalars @xmath7 and @xmath8 , and a neutral cp - odd pseudoscalar @xmath9 .
the higgs potential is constrained by supersymmetry such that all tree - level higgs boson masses and couplings are determined by just two independent parameters , commonly chosen to be the mass of the cp - odd pseudoscalar ( @xmath10 ) and the ratio of vacuum expectation values of the neutral higgs fields ( @xmath11 ) . in the standard model ,
gluon fusion is the dominant process for producing a pair of higgs bosons via triangle and box diagrams with internal top quarks and bottom quarks @xcite .
the yukawa coupling of the higgs boson to bottom quarks in the standard model is proportional to @xmath12 , where @xmath13 is the vacuum - expectation value of the higgs field , and is hence very weak .
thus , at the fermilab tevatron and cern large hadron collider(lhc ) , the rate for inclusive higgs pair production from @xmath0 quark fusion is small @xcite .
however , it can become significant in supersymmetric models with large @xmath1 since the yukawa couplings of the higgs bosons to the @xmath0 quark are enhanced by @xmath14 . in the minimal supersymmetric model ( mssm ) , the production rate of pairs of the lightest neutral higgs boson ( @xmath15 ) from @xmath16 fusion is larger than the rate from the gluon - gluon initial state for @xmath17 and moderate values of @xmath10 ( @xmath18 ) , while the rate for pair production of the heavier neutral higgs bosons ( @xmath19 and @xmath20 ) from @xmath21 fusion is highly suppressed relative to the gluon - gluon production mechanism at small tan@xmath22@xcite .
this makes it of interest to evaluate higher order corrections to the rates .
one of the most powerful ways to distinguish between the higgs boson of the standard model and those of the mssm is to measure the trilinear neutral higgs boson couplings .
this measurement is extremely challenging in most scenarios @xcite .
the enhanced higgs pair production rates from bottom quark fusion when @xmath1 is large provide discovery potential for determining the trilinear higgs couplings in the mssm . the rate for higgs boson production from bottom quark fusion@xcite has been computed to nnlo@xcite and the electroweak and susy qcd corrections included to one - loop@xcite .
also , the rate for higgs production in association with a @xmath0 quark has been computed to nlo@xcite including susy qcd corrections@xcite . in this paper
, we present a complete next - to - leading order ( nlo ) susy qcd calculation for higgs pair production via bottom quark fusion .
the leading order ( lo ) process is @xmath23 ( where @xmath24 ) and the nlo cross section includes both @xmath25 and @xmath26 corrections , where @xmath27 @xcite .
the subprocess @xmath28 contributes @xmath26 corrections to the leading order cross section for @xmath23 .
theoretical predictions depend on the number of @xmath0 quarks tagged . here
, we consider only inclusive processes in which there are no tagged @xmath0 quarks .
the nlo pure qcd corrections have been computed in ref .
the focus of this work is the inclusion of the @xmath25 susy qcd ( sqcd ) corrections , which consist of squark and gluino loop contributions .
we also present a detailed study of the effects of the susy parameters on the production rate in the mssm and msugra models . in section
ii , we review the leading - order cross section for @xmath29 from @xmath30 fusion . in section iii , we provide the complete next - to - leading order ( nlo ) sqcd corrections from gluino and squark loops for @xmath31 production .
numerical results are given in section iv and conclusions are drawn in section v. in addition , there are three appendices .
appendix a defines the scalar integrals @xcite used in our computation , appendix b contains the coefficients used in computing the virtual sqcd corrections and appendix c presents the yukawa couplings , higgs trilinear couplings and higgs - squark - squark couplings in the mssm .
the leading order ( lo ) inclusive cross section for @xmath33 via @xmath34 is @xmath35 \hat{\sigma}_{lo}(s , t , u)(b\bar{b } \to \phi\phi ) \label{sigma : lo}\end{aligned}\ ] ] where @xmath36 and @xmath37 are the lo parton distribution functions for bottom quarks in the proton , @xmath38 is the parton level cross section for @xmath39 and @xmath40 are the mandelstam variables . fig . [
fig : tree ] shows the tree level feynman diagrams for @xmath39 in the mssm .
= 3.5 in we assign momenta to the initial and the final state partons with @xmath41 and @xmath42 .
the @xmath43 yukawa coupling , @xmath44 , and the @xmath45 and @xmath46 trilinear couplings , @xmath47 and @xmath48 , can be found in appendix c. the lowest order rate can be expressed in terms of spinor structures : @xmath49 with @xmath50 . following the simplified acot prescription @xcite , we take @xmath51 everywhere except in the yukawa couplings .
the tree level amplitudes of the @xmath52 and @xmath53 channels are : @xmath54 where @xmath55 are color indices .
the corresponding spin- and color - averaged matrix elements squared , including interferences terms , are @xmath56 the lo element is then , @xmath57 finally , the parton level cross section for inclusive @xmath34 production is @xmath58 where @xmath59 denotes the integral over the two - body phase space and the factor of @xmath60 is from the identical particles in the final state .
the parton level nlo cross section is @xmath61 \nonumber \\ \delta\hat{\sigma}_{\rm nlo}(x_1,x_2 , \mu ) & \equiv & \delta\hat{\sigma}_{\rm qcd}(x_1 , x_2 , \mu ) + \delta\hat{\sigma}_{\rm sqcd}(x_1 , x_2 , \mu)\ , , \label{sigma : pnlo}\end{aligned}\ ] ] where @xmath62 is the leading order ( born ) cross section and @xmath63 is the next - to - leading order correction to the born cross section , @xmath64 are the momentum fractions of the partons and @xmath65 is the renormalization scale .
the nlo correction @xmath66 contains both the gluon qcd correction , @xmath67 , and the gluino - squark sqcd correction , @xmath68 .
the gluon qcd correction includes the @xmath69 corrections , which contain virtual and real gluon emission contributions , as well as the @xmath70 corrections from the @xmath71 subprocess .
the gluino - squark sqcd correction contains @xmath69 gluino - sbottom loop contributions . unlike down - type quarks , which only couple to the down - type higgs at tree level , the down - type squarks also couple to the up - type higgs boson
this leads to mixing in the sbottom mass matrix in the @xmath72 basis , @xmath73 where @xmath74 are the mass eigenstates .
the mass eigenstates @xmath75 and @xmath76 ( @xmath77 ) are defined in terms of a mixing angle , @xmath78 the one - loop virtual gluino - sbottom feynman diagrams for @xmath34 are shown in fig .
[ fig : oneloop ] .
= 5.5 in the amplitudes corresponding to each diagram in fig .
[ fig : oneloop ] are computed analytically and all tensor integrals are reduced to linear combinations of one - loop scalar functions .
we sum all of the virtual amplitudes to obtain the contribution to the matrix element from gluino - sbottom loops in terms of the standard matrix elements of eqs .
[ sdef ] and [ smes ] , @xmath79 where @xmath80 corresponds to the numbering of the diagrams in fig .
[ fig : oneloop ] .
the analytic expressions for each coefficient @xmath81 can be found in appendix b. for simplification , we define : @xmath82 the one - loop sqcd matrix element is then @xmath83 only ultraviolet ( uv ) divergences occur in the susy qcd corrections from the massive gluino and squark loops .
these uv divergences are removed by the renormalization of the bottom quark wavefunction and propagator and the renormalization of the bottom quark mass in the yukawa coupling @xcite .
we use the on - shell renormalization scheme of ref .
@xcite and define the @xmath0 quark self energy as , @xmath84 which gives the renormalized propagator , @xmath85 the on - shell renormalization condition requires , @xmath86 computing the bottom quark self energy from gluino - squark loops and ignoring contributions suppressed by powers of @xmath87 , we find : @xmath88 \nonumber \\
\sigma_s^b(p^2 ) & = & -\frac{\alpha_s}{3\pi}\biggl({m_{\tilde g}\over m_b}\biggr ) \sin2\theta_{\tilde b } \left [ b_0(p , m_{\tilde g } , m_{\tilde b_2 } ) - b_0(p , m_{\tilde g } , m_{\tilde b_1})\right]\ , , \end{aligned}\ ] ] which yields , @xmath89 the coupling of the @xmath0 squark to the up - type higgs doublet induces a modification of the tree - level relation between the bottom quark mass and its yukawa coupling . at large @xmath1
, we absorb this modification by redefining the bottom quark mass occuring in the yukawa coupling,@xcite @xmath90 where @xmath91 and the auxiliary function is defined as , @xmath92 the contributions to the bottom yukawa couplings which are enhanced at large @xmath1 can be included to all orders by making the following replacements @xcite : @xmath93 we use these effective yukawa couplings in our nlo computation . to avoid double - counting ,
we add additional mass counterterms , @xmath94 for @xmath7 , @xmath8 and @xmath9 production , respectively .
the counterterms arise from the @xmath0 quark wavefunction renormalization ( a factor of @xmath95 for each external @xmath0 quark ) , the @xmath0 mass renormalization in the @xmath96 couplings ( this occurs twice in the @xmath97 and @xmath53 channel diagrams , and once in the @xmath98channel diagram ) , and a factor of @xmath99 from the mass renormalization on the internal @xmath0 quark propagators shown in fig .
[ fg : counter ] : @xmath100 = 3.5 in the complete one - loop susy qcd contribution is , @xmath101 and is of course finite . finally , the sqcd contribution to the matrix element squared is , @xmath102 with the spin and color averaged result , @xmath103 s \right\}\ , .
\end{aligned}\ ] ]
in this section , we present the next - to - leading - order inclusive cross sections for the production of a pair of neutral higgs bosons via bottom quark fusion in the mssm and msugra models at the cern lhc . as in our previous paper @xcite , we use the lowest order cteq6l1 parton distribution functions ( pdfs ) @xcite with a factorization scale @xmath104 and the leading - order evolution of the strong coupling @xmath105 with a renormalization scale @xmath106 to calculate the lo cross sections and use the cteq6 m pdfs with the next - to - leading - order evolution of @xmath105 to evaluate the nlo inclusive cross sections . for simplification
, we use the same renormalization and factorization scales @xmath107 .
we evaluate the bottom quark mass occuring in the @xmath108 yukawa couplings using the @xmath109 mass , @xmath110 , with a two - loop heavy quark running mass with @xmath111 and the nlo evolution of the strong coupling constant , modified to decouple the effects of the top quark@xcite .
the higgs couplings are in appendix c @xcite and we compute the higgs boson masses to one - loop accuracy @xcite .
= 6.5 in in fig .
[ fig : mssm1 ] , we show the lo and nlo cross sections versus the pseudoscalar higgs mass @xmath10 .
we assume @xmath112 and compute the @xmath0 squark masses and mixing angles from eqa .
[ bsmass ] and [ bsmixing ] .
our nlo cross sections include the @xmath25 corrections from the @xmath16 initial state and the @xmath26 corrections from the @xmath113 initial state , along with the sqcd corrections from gluino - sbottom loops .
we show our results with @xmath114 ( red ) and @xmath115 ( blue ) at @xmath116 gev . to compare with @xmath117 fusion
, we also plot the cross section from the @xmath117 initial state ( dot ) @xcite .
we note that the cross sections for higgs pair production in the mssm are significantly larger than in the standard model , due to enhancements at large @xmath1 .
the resonant enhancements due to @xmath98 channel scalar exchange are clearly visible in the @xmath118 and @xmath119 curves . at @xmath115 ,
the cross section for pair of the lightest neutral higgs boson ( @xmath15 ) from @xmath16 fusion is much larger than the cross section from the gluon - gluon initial state , while the rate for pair production of the heavier neutral higgs bosons ( @xmath19 and @xmath20 ) from @xmath16 fusion is highly suppressed relative to the gluon - gluon production .
but at @xmath114 , gluon fusion dominates pair of the lightest neutral higgs boson ( @xmath15 ) production and is comparible with @xmath16 production for the heavier neutral higgs bosons ( @xmath19 and @xmath20 ) production .
= 3.in figs .
[ fig : mssm2 ] and fig .
[ fig : mssm3 ] show the nlo cross sections versus @xmath10 with @xmath116 gev and @xmath114 .
we present the nlo cross section with only gluon qcd corrections ( dash - dot - dot , green ) , nlo cross section with only gluino - sbottom sqcd corrections ( dot , blue),and the complete nlo cross section with qcd and sqcd corrections together ( dash , red ) .
we note : * the nlo sqcd correction to @xmath120 is small .
the dominant contribution to the nlo correction to light higgs pair production is from the pure qcd contribution . * for heavy higgs pair and pseudoscalar higgs pair production ,
the susy qcd corrections become dominant .
the pure gluon nlo contribution is much smaller in magnitude than the contribution from sqcd . in fig .
[ fig : mssm4 ] , we plot the ratio of the nlo sqcd correction normalized to the born cross section , @xmath121 , with @xmath122 gev , and @xmath115 ( dash - dot - dot , green ) , @xmath123 ( dash , blue ) and @xmath124 ( solid , red ) . in the limit of large squark masses , @xmath125 approaches a common non - zero constant for @xmath19 and @xmath20 production .
light higgs pair production , however , decouples for large @xmath10 and large susy masses .
this decoupling behaviour is also seen in the decay @xmath126
@xcite and the production process @xmath127@xcite . in fig .
[ fig : mssm5 ] , we fix @xmath10 and all squark masses to be @xmath128 gev and plot @xmath125 versus the gluino mass @xmath129 with @xmath130 ( dash - dot - dot , green ) , @xmath123 ( dash , blue ) and @xmath124 ( solid , red ) . this figure does not demonstrate a decoupling behaviour . in this model , supersymmetry is assumed to be broken in a hidden sector consisting of fields that interact with the usual particles and their superpartners only via gravity .
supersymmetry breaking is communicated to the visible sector via gravitational interactions . within the msugra framework
, it is assumed that at some high scale ( frequently taken to be @xmath131 ) all scalar fields have a common susy breaking mass @xmath132 , all gauginos have a common mass @xmath133 , and all soft susy breaking scalar trilinear couplings have a common value @xmath134 .
electroweak symmetry breaking is assumed to occur radiatively .
this fixes the magnitude of the superpotential parameter @xmath135 .
the soft susy breaking bilinear higgs boson mass parameter can be eliminated in favour of @xmath1 , so that the model is completely specified by the parameter set : @xmath136 all the sparticle masses and couplings required for phenomenological analysis can be obtained via renormalization group evolution between the scale of grand unification and the weak scale .
we show the lo ( dot ) and nlo ( solid ) cross sections versus @xmath132 with @xmath137 for @xmath115 in fig .
[ fig : msugra1 ] and @xmath114 in fig .
[ fig : msugra2 ] .
we plot six curves in each frame , three lo cross sections ( dot ) and three nlo cross sections ( solid ) .
the nlo curves include both the pure qcd and the sqcd contributions . as shown in the graphs , blue lines are cross sections with @xmath138 gev , @xmath139 , green lines have @xmath138 gev , @xmath140 , and red lines are for @xmath141 gev , @xmath139 . comparing figs .
[ fig : msugra1 ] and [ fig : msugra2 ] we see a strong dependence on @xmath1 .
* when @xmath115 , flipping the sign of @xmath135 has little effect on either the lo or the nlo cross sections with @xmath138 gev .
as @xmath1 increases to 50 , we notice that flipping the sign of @xmath135 has a large effect when @xmath142 gev as shown in fig .
[ fig : msugra2 ] . * for large @xmath132 , the cross sections approach a common value , independent of @xmath133 and @xmath135 .
in this paper we presented the complete @xmath25 susy qcd corrections to neutral higgs pair production from bottom quark fusion in the minimal supersymmetric model and the minimal supergravity model at the cern lhc .
the effects of the susy qcd corrections from sbottom and gluino loops are significant in some regions of parameter space and the decoupling of the sqcd effects is only recovered for light higgs pair production in the large @xmath10 limit .
we are grateful to jianwei qiu and chris jackson for beneficial discussions .
this research was supported in part by the u.s .
department of energy under grants no .
de - ac02 - 76ch1 - 886 , no .
de - fg02 - 04er41305 , no .
de - fg02 - 03er46040 and de - ac02 - 98ch10886 .
sd thanks the slac theory group for their hospitality , where this work was completed .
@xmath143 } \nonumber \\ { i\over 16\pi^2}\not{p } b_{11}(p , m_1,m_2 ) & = & \int{d^nk\over ( 2\pi)^n } \frac{\not k}{(k^2-m_1 ^ 2 ) [ ( k+p)^2-m_2 ^ 2 ] } \nonumber \\ { i\over 16\pi^2}c_{0}(p_1 , p_2 , m_1 , m_2,m_3 ) & = & \int { d^nk\over ( 2\pi)^n } \frac{1}{(k^2-m_1 ^ 2 ) [ ( k+p_1)^2-m_2 ^ 2][(k+p_1+p_2)^2 -m_3 ^ 2 ] } \nonumber\end{aligned}\ ] ] @xmath144[(k+p_1+p_2)^2 -m_3 ^ 2 ] } \nonumber \end{aligned}\ ] ] @xmath145[(k+p_1+p_2)^2 -m_3 ^ 2][(k+p_1+p_2+p_3)^2-m_4 ^ 2 ] } \nonumber \\ & & \nonumber \\ & & d_{123}(p_1 , p_2,p_3 , m_1 , m_2,m_3,m_4 ) \nonumber \\ & & = d_{12}(p_1 , p_2,p_3 , m_1 , m_2,m_3,m_4 ) - d_{13}(p_1 , p_2,p_3 , m_1 , m_2,m_3,m_4 ) \nonumber \end{aligned}\ ] ]
@xmath148 \nonumber \\ x_{12 } & = & \frac{1}{8\pi^2 } \left [ b_1(-p_1+p_3 , m_{\tilde{g } } , m_{\tilde b_1 } ) \cos^2\theta_{\tilde b } + b_1(-p_1+p_3 , m_{\tilde{g } } , m_{\tilde b_2 } ) \sin^2\theta_{\tilde b}\right ] \nonumber \\ x_{13 } & = & -\frac{1}{8\pi^2 } \frac{g_{\phi bb}^2}{t } m_{\tilde g } \left[b_{0}(-p_1+p_3 , m_{\tilde g } , m_{\tilde b_1 } ) - b_{0}(-p_1+p_3 , m_{\tilde g } , m_{\tilde b_2 } ) \right]\sin\theta_{\tilde b } \cos\theta_{\tilde b } \nonumber \\ x_{14 } & = & x_{13 } \nonumber \\ \nonumber \\ x_{2i } & = & x_{1i } ( t \leftrightarrow u , p_3 \leftrightarrow p_4 ) \quad i = 1,2,3,4 \nonumber \\ \nonumber \\ x_{31 } & = & \frac{1}{8\pi^2}\frac{g_{\phi11 } } { g_{\phi bb } } m_{\tilde g } c_0(-p_1,p_3 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_1 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b}\nonumber \\ & + & \frac{1}{8\pi^2}\frac{g_{\phi12 } } { g_{\phi bb } } m_{\tilde g } c_0(-p_1,p_3 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2 } ) \cos^2\theta_{\tilde b } \nonumber \\ & - & \frac{1}{8\pi^2}\frac{g_{\phi12 } } { g_{\phi bb } } m_{\tilde g } c_0(-p_1,p_3 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1 } ) \sin^2\theta_{\tilde b } \nonumber \\ & - & \frac{1}{8\pi^2}\frac{g_{\phi22 } } { g_{\phi bb } } m_{\tilde g } c_0(-p_1,p_3 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_2 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b } \nonumber \\ x_{32 } & = & \frac{1}{8\pi^2}\frac{g_{\phi11 } } { g_{\phi bb } } m_{\tilde g } c_0(-p_1,p_3 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_1 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b}\nonumber \\ & - & \frac{1}{8\pi^2}\frac{g_{\phi12 } } { g_{\phi bb } } m_{\tilde g } c_0(-p_1,p_3 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2 } ) \sin^2\theta_{\tilde b } \nonumber \\ & + & \frac{1}{8\pi^2}\frac{g_{\phi12 } } { g_{\phi bb } } m_{\tilde g } c_0(-p_1,p_3 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1 } ) \cos^2\theta_{\tilde b } \nonumber \\ & - & \frac{1}{8\pi^2}\frac{g_{\phi22 } } { g_{\phi bb } } m_{\tilde g } c_0(-p_1,p_3 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_2 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b } \nonumber \\ x_{33 } & = & -\frac{1}{8\pi^2 } g_{\phi bb } g_{\phi11 } c_{12}(-p_1,p_3 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_1 } ) \cos^2\theta_{\tilde b } \nonumber \\ & + & \frac{1}{8\pi^2 } g_{\phi bb } g_{\phi12 } c_{12}(-p_1,p_3 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b } \nonumber \\ & + & \frac{1}{8\pi^2 } g_{\phi bb } g_{\phi12 } c_{12}(-p_1,p_3 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b } \nonumber \\ & - & \frac{1}{8\pi^2 } g_{\phi bb } g_{\phi22 } c_{12}(-p_1,p_3 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_2 } ) \sin^2\theta_{\tilde b } \nonumber \\ x_{34 } & = & -\frac{1}{8\pi^2 } g_{\phi bb } g_{\phi11 } c_{12}(-p_1,p_3 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_1 } ) \sin^2\theta_{\tilde b } \nonumber \\ & - & \frac{1}{8\pi^2 } g_{\phi bb } g_{\phi12 } c_{12}(-p_1,p_3 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b } \nonumber \\ & - & \frac{1}{8\pi^2 } g_{\phi bb } g_{\phi12 } c_{12}(-p_1,p_3 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b } \nonumber \\ & - & \frac{1}{8\pi^2 } g_{\phi bb } g_{\phi22 } c_{12}(-p_1,p_3 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_2 } ) \cos^2\theta_{\tilde b } \nonumber \\ \nonumber \\ x_{4i } & = & x_{3i } ( t \leftrightarrow u , p_3 \leftrightarrow p_4 ) \quad i = 1,2,3,4 \nonumber \\ \nonumber \\ x_{51 } & = & \frac{1}{8\pi^2}\frac{g_{\phi11 } } { g_{\phi bb } } m_{\tilde g } c_0(p_2,-p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_1 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b}\nonumber \\ & - & \frac{1}{8\pi^2}\frac{g_{\phi12 } } { g_{\phi bb } } m_{\tilde g } c_0(p_2,-p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1 } ) \sin^2\theta_{\tilde b } \nonumber \\ & + & \frac{1}{8\pi^2}\frac{g_{\phi12 } } { g_{\phi bb } } m_{\tilde g } c_0(p_2,-p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2 } ) \cos^2\theta_{\tilde b } \nonumber \\ & - & \frac{1}{8\pi^2}\frac{g_{\phi22 } } { g_{\phi bb } } m_{\tilde g } c_0(p_2,-p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_2 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b}\nonumber \\ x_{52 } & = & \frac{1}{8\pi^2}\frac{g_{\phi11 } } { g_{\phi bb } } m_{\tilde g } c_0(p_2,-p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_1 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b}\nonumber \\ & + & \frac{1}{8\pi^2}\frac{g_{\phi12 } } { g_{\phi bb } } m_{\tilde g } c_0(p_2,-p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1 } ) \cos^2\theta_{\tilde b } \nonumber \\ & - & \frac{1}{8\pi^2}\frac{g_{\phi12 } } { g_{\phi bb } } m_{\tilde g } c_0(p_2,-p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2 } ) \sin^2\theta_{\tilde b } \nonumber \\ & - & \frac{1}{8\pi^2}\frac{g_{\phi22 } } { g_{\phi bb } } m_{\tilde g } c_0(p_2,-p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_2 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b}\nonumber \\ x_{53 } & = & -\frac{1}{8\pi^2 } g_{\phi bb } g_{\phi11 } c_{12}(p_2,-p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_1 } ) \sin^2\theta_{\tilde b } \nonumber \\ & - & \frac{1}{8\pi^2 } g_{\phi bb } g_{\phi12 } c_{12}(p_2,-p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b } \nonumber \\ & - & \frac{1}{8\pi^2 } g_{\phi bb } g_{\phi12 } c_{12}(p_2,-p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b } \nonumber \\ & - & \frac{1}{8\pi^2 } g_{\phi bb } g_{\phi22 } c_{12}(p_2,-p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_2 } ) \cos^2\theta_{\tilde b } \nonumber \\ x_{54 } & = & -\frac{1}{8\pi^2 } g_{\phi bb } g_{\phi11 } c_{12}(p_2,-p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_1 } ) \cos^2\theta_{\tilde b } \nonumber \\ & + & \frac{1}{8\pi^2 } g_{\phi bb } g_{\phi12 } c_{12}(p_2,-p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b } \nonumber \\ & + & \frac{1}{8\pi^2 } g_{\phi bb } g_{\phi12 } c_{12}(p_2,-p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b } \nonumber \\ & - & \frac{1}{8\pi^2 } g_{\phi bb } g_{\phi22 } c_{12}(p_2,-p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_2 } ) \sin^2\theta_{\tilde b } \nonumber \\ \nonumber \\ x_{6i } & = & x_{5i } ( t \leftrightarrow u , p_3 \leftrightarrow p_4 ) \quad i = 1,2,3,4 \nonumber \\ \nonumber \\ x_{71 } & = & \frac{1}{8\pi^2 } \frac{g_{\phi11}^2}{g_{\phi bb}^2 } t d_{123}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_1},m_{\tilde b_1 } ) \cos^2\theta_{\tilde b } \nonumber \\ & - & \frac{1}{8\pi^2 } \frac{g_{\phi11 } g_{\phi12}}{g_{\phi bb}^2 } t d_{123}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_1},m_{\tilde b_2 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b } \nonumber \\ & + & \frac{1}{8\pi^2 } \frac{g_{\phi12}^2}{g_{\phi bb}^2 } t d_{123}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2},m_{\tilde b_1 } ) \cos^2\theta_{\tilde b}\nonumber \\ & - & \frac{1}{8\pi^2 } \frac{g_{\phi22 } g_{\phi12}}{g_{\phi bb}^2 } t d_{123}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2},m_{\tilde b_2 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b } \nonumber \\ & - & \frac{1}{8\pi^2 } \frac{g_{\phi11 } g_{\phi12}}{g_{\phi bb}^2 } t d_{123}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1},m_{\tilde b_1 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b } \nonumber \\ & + & \frac{1}{8\pi^2 } \frac{g_{\phi12}^2}{g_{\phi bb}^2 } t d_{123}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1},m_{\tilde b_2 } ) \sin^2\theta_{\tilde b } \nonumber \\ & - & \frac{1}{8\pi^2 } \frac{g_{\phi22 } g_{\phi12}}{g_{\phi bb}^2 } t d_{123}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_2},m_{\tilde b_1 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b}\nonumber \\ & + & \frac{1}{8\pi^2 } \frac{g_{\phi22}^2}{g_{\phi bb}^2 } t d_{123}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_2},m_{\tilde b_2 } ) \sin^2\theta_{\tilde b}\nonumber \\ x_{72 } & = & \frac{1}{8\pi^2 } \frac{g_{\phi11}^2}{g_{\phi bb}^2 } t d_{123}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_1},m_{\tilde b_1 } ) \sin^2\theta_{\tilde b } \nonumber \\ & + & \frac{1}{8\pi^2 } \frac{g_{\phi11 } g_{\phi12}}{g_{\phi bb}^2 } t d_{123}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_1},m_{\tilde b_2 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b}\nonumber \\ & + & \frac{1}{8\pi^2 } \frac{g_{\phi12}^2}{g_{\phi bb}^2 } t d_{123}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2},m_{\tilde b_1 } ) \sin^2\theta_{\tilde b}\nonumber \\ & + & \frac{1}{8\pi^2 } \frac{g_{\phi22 } g_{\phi12}}{g_{\phi bb}^2 } t d_{123}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2},m_{\tilde b_2 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b } \nonumber \\ & + & \frac{1}{8\pi^2 } \frac{g_{\phi11 } g_{\phi12}}{g_{\phi bb}^2 } t d_{123}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1},m_{\tilde b_1 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b } \nonumber \\ & + & \frac{1}{8\pi^2 } \frac{g_{\phi12}^2}{g_{\phi bb}^2 } t d_{123}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1},m_{\tilde b_2 } ) \cos^2\theta_{\tilde b}\nonumber \\ & + & \frac{1}{8\pi^2 } \frac{g_{\phi22 } g_{\phi12}}{g_{\phi bb}^2 } t d_{123}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_2},m_{\tilde b_1 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b } \nonumber \\ & + & \frac{1}{8\pi^2 } \frac{g_{\phi22}^2}{g_{\phi bb}^2 } t d_{123}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_2},m_{\tilde b_2 } ) \cos^2\theta_{\tilde b}\nonumber \\ x_{73 } & = & -\frac{1}{8\pi^2 } g^2_{\phi11 } m_{\tilde g } d_{0}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_1},m_{\tilde b_1 } ) \sin\theta_{\tilde b}\cos\theta_{\tilde b}\nonumber \\ & - & \frac{1}{8\pi^2 } g_{\phi11 } g_{\phi12 } m_{\tilde g } d_{0}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_1},m_{\tilde b_2 } ) \cos^2\theta_{\tilde b } \nonumber \\ & - & \frac{1}{8\pi^2 } g^2_{\phi12 } m_{\tilde g } d_{0}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2},m_{\tilde b_1 } ) \sin\theta_{\tilde b}\cos\theta_{\tilde b}\nonumber \\ & - & \frac{1}{8\pi^2 } g_{\phi22 } g_{\phi12 } m_{\tilde g } d_{0}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2},m_{\tilde b_2 } ) \cos^2\theta_{\tilde b } \nonumber \\ & + & \frac{1}{8\pi^2 } g_{\phi11 } g_{\phi12 } m_{\tilde g } d_{0}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1},m_{\tilde b_1 } ) \sin^2\theta_{\tilde b } \nonumber \\ & + & \frac{1}{8\pi^2 } g^2_{\phi12 } m_{\tilde g } d_{0}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1},m_{\tilde b_2 } ) \sin\theta_{\tilde b}\cos\theta_{\tilde b}\nonumber \\ & + & \frac{1}{8\pi^2 } g_{\phi22 } g_{\phi12 } m_{\tilde g } d_{0}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_2},m_{\tilde b_1 } ) \sin^2\theta_{\tilde b } \nonumber \\ & + & \frac{1}{8\pi^2 } g^2_{\phi22 } m_{\tilde g } d_{0}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_2},m_{\tilde b_2 } ) \sin\theta_{\tilde b}\cos\theta_{\tilde b}\nonumber \\ x_{74 } & = & -\frac{1}{8\pi^2 } g^2_{\phi11 } m_{\tilde g } d_{0}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_1},m_{\tilde b_1 } ) \sin\theta_{\tilde b}\cos\theta_{\tilde b}\nonumber \\ & + & \frac{1}{8\pi^2 } g_{\phi11 } g_{\phi12 } m_{\tilde g } d_{0}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_1},m_{\tilde b_2 } ) \sin^2\theta_{\tilde b } \nonumber \\ & - & \frac{1}{8\pi^2 } g^2_{\phi12 } m_{\tilde g } d_{0}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2},m_{\tilde b_1 } ) \sin\theta_{\tilde b}\cos\theta_{\tilde b}\nonumber \\ & + & \frac{1}{8\pi^2 } g_{\phi22 } g_{\phi12 } m_{\tilde g } d_{0}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2},m_{\tilde b_2 } ) \sin^2\theta_{\tilde b } \nonumber \\ & - & \frac{1}{8\pi^2 } g_{\phi11 } g_{\phi12 } m_{\tilde g } d_{0}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1},m_{\tilde b_1 } ) \cos^2\theta_{\tilde b } \nonumber \\ & + & \frac{1}{8\pi^2 } g^2_{\phi12 } m_{\tilde g } d_{0}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1},m_{\tilde b_2 } ) \sin\theta_{\tilde b}\cos\theta_{\tilde b}\nonumber
\\ & - & \frac{1}{8\pi^2 } g_{\phi22 } g_{\phi12 } m_{\tilde g } d_{0}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_2},m_{\tilde b_1 } ) \cos^2\theta_{\tilde b } \nonumber \\ & + & \frac{1}{8\pi^2 } g^2_{\phi22 } m_{\tilde g } d_{0}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_2},m_{\tilde b_2 } ) \sin\theta_{\tilde b}\cos\theta_{\tilde b}\nonumber \\ \nonumber \\ x_{8i } & = & x_{7i } ( t \leftrightarrow u , p_3 \leftrightarrow p_4 ) \quad i = 1,2,3,4 \nonumber \\ \nonumber \\ x_{93 } & = & -\frac{1}{8\pi^2 } g_{h11}g_{h\phi\phi}\frac { m_{\tilde g}}{(s - m_h^2 ) + i m_h \gamma_h } c_{0}(-p_1,p_1+p_2 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_1 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b}\nonumber \\ & - & \frac{1}{8\pi^2 } g_{h12}g_{h\phi\phi}\frac { m_{\tilde g}}{(s - m_h^2 ) + i m_h \gamma_h } c_{0}(-p_1,p_1+p_2 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2 } ) \cos^2\theta_{\tilde b}\nonumber \\ & + & \frac{1}{8\pi^2 } g_{h12}g_{h\phi\phi}\frac { m_{\tilde g}}{(s - m_h^2 ) + i m_h \gamma_h } c_{0}(-p_1,p_1+p_2 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1 } ) \sin^2\theta_{\tilde b}\nonumber \\ &
+ & \frac{1}{8\pi^2 } g_{h22}g_{h\phi\phi}\frac { m_{\tilde g}}{(s - m_h^2 ) + i m_h \gamma_h } c_{0}(-p_1,p_1+p_2 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_2 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b}\nonumber \\ x_{94 } & = & -\frac{1}{8\pi^2 } g_{h11}g_{h\phi\phi}\frac { m_{\tilde g}}{(s - m_h^2 ) + i m_h \gamma_h } c_{0}(-p_1,p_1+p_2 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_1 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b}\nonumber \\ & + & \frac{1}{8\pi^2 } g_{h12}g_{h\phi\phi}\frac { m_{\tilde g}}{(s - m_h^2 ) + i m_h \gamma_h } c_{0}(-p_1,p_1+p_2 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2 } ) \sin^2\theta_{\tilde b } \nonumber \\ & - & \frac{1}{8\pi^2 } g_{h12}g_{h\phi\phi}\frac { m_{\tilde g}}{(s - m_h^2 ) +
i m_h \gamma_h } c_{0}(-p_1,p_1+p_2 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1 } ) \cos^2\theta_{\tilde b } \nonumber \\ & + & \frac{1}{8\pi^2 } g_{h22}g_{h\phi\phi}\frac { m_{\tilde g}}{(s - m_h^2 ) + i
m_h \gamma_h } c_{0}(-p_1,p_1+p_2 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_2 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b}\nonumber \\ \nonumber \\ x_{10i } & = & x_{9i } ( g_{hjk } \leftrightarrow g_{hjk } , g_{h\phi\phi } \leftrightarrow g_{h\phi\phi } ) \quad i=3 , 4 ; \quad j , k = 1 , 2 \nonumber \\ \nonumber \\\end{aligned}\ ] ] @xmath150 \nonumber \\ x_{12 } & = & \frac{1}{8\pi^2 } \left [ b_1(-p_1+p_3 , m_{\tilde{g } } , m_{\tilde b_1 } ) \cos^2\theta_{\tilde b } + b_1(-p_1+p_3 , m_{\tilde{g } } , m_{\tilde b_2 } ) \sin^2\theta_{\tilde b}\right ] \nonumber \\ x_{13 } & = & \frac{1}{8\pi^2 } \frac{g_{hbb}^2}{t } m_{\tilde g } \left[b_{0}(-p_1+p_3 , m_{\tilde g } , m_{\tilde b_1 } ) - b_{0}(-p_1+p_3 , m_{\tilde g } , m_{\tilde b_2 } ) \right]\sin\theta_{\tilde b } \cos\theta_{\tilde b } \nonumber \\ x_{14 } & = & x_{13 } \nonumber \\ \nonumber \\ x_{2i } & = & x_{1i } ( t \leftrightarrow u , p_3 \leftrightarrow p_4 ) \quad i = 1,2,3,4 \nonumber \\ \nonumber \\
x_{31 } & = & \frac{1}{8\pi^2}\frac{g_{a12 } } { g_{abb } } m_{\tilde g } c_0(-p_1,p_3 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2 } ) \cos^2\theta_{\tilde b } \nonumber \\ & + & \frac{1}{8\pi^2}\frac{g_{a12 } } { g_{abb } } m_{\tilde g } c_0(-p_1,p_3 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1 } ) \sin^2\theta_{\tilde b } \nonumber \\
x_{32 } & = & \frac{1}{8\pi^2}\frac{g_{a12 } } { g_{abb } } m_{\tilde g } c_0(-p_1,p_3 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2 } ) \sin^2\theta_{\tilde b } \nonumber \\ & + & \frac{1}{8\pi^2}\frac{g_{a12 } } { g_{abb } } m_{\tilde g } c_0(-p_1,p_3 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1 } ) \cos^2\theta_{\tilde b } \nonumber \\
x_{33 } & = & -\frac{1}{8\pi^2 } g_{abb } g_{a12 } c_{12}(-p_1,p_3 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b } \nonumber \\ & + & \frac{1}{8\pi^2 } g_{abb } g_{a12 } c_{12}(-p_1,p_3 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b } \nonumber \\
x_{34 } & = & -\frac{1}{8\pi^2 } g_{abb } g_{a12 } c_{12}(-p_1,p_3 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b } \nonumber \\ & + & \frac{1}{8\pi^2 } g_{abb } g_{a12 } c_{12}(-p_1,p_3 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b } \nonumber
\\ \nonumber \\ x_{4i } & = & x_{3i } ( t \leftrightarrow u , p_3 \leftrightarrow p_4 ) \quad i = 1,2,3,4 \nonumber \\ \nonumber \\ x_{51 } & = & \frac{1}{8\pi^2}\frac{g_{a12 } } { g_{abb } } m_{\tilde g } c_0(p_2,-p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1 } ) \sin^2\theta_{\tilde b } \nonumber \\ & + & \frac{1}{8\pi^2}\frac{g_{a12 } } { g_{abb } } m_{\tilde g } c_0(p_2,-p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2 } ) \cos^2\theta_{\tilde b } \nonumber \\ x_{52 } & = & \frac{1}{8\pi^2}\frac{g_{a12 } } { g_{abb } } m_{\tilde g } c_0(p_2,-p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1 } ) \cos^2\theta_{\tilde b } \nonumber \\ & + & \frac{1}{8\pi^2}\frac{g_{a12 } } { g_{abb } } m_{\tilde g } c_0(p_2,-p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2 } ) \sin^2\theta_{\tilde b } \nonumber \\ x_{53 } & = & \frac{1}{8\pi^2 } g_{abb } g_{a12 } c_{12}(p_2,-p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b } \nonumber \\ & - & \frac{1}{8\pi^2 } g_{abb } g_{a12 } c_{12}(p_2,-p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b } \nonumber \\ x_{54 } & = & \frac{1}{8\pi^2 } g_{abb } g_{a12 } c_{12}(p_2,-p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b } \nonumber \\ & - & \frac{1}{8\pi^2 } g_{abb } g_{a12 } c_{12}(p_2,-p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b } \nonumber
\\ \nonumber \\ x_{6i } & = & x_{5i } ( t \leftrightarrow u , p_3 \leftrightarrow p_4 ) \quad i = 1,2,3,4 \nonumber \\ \nonumber \\ x_{71 } & = & \frac{1}{8\pi^2 } \frac{g_{a12}^2}{g_{abb}^2 } t d_{123}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2},m_{\tilde b_1 } ) \cos^2\theta_{\tilde b } \nonumber \\ & + & \frac{1}{8\pi^2 } \frac{g_{a12}^2}{g_{abb}^2 } t d_{123}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1},m_{\tilde b_2 } ) \sin^2\theta_{\tilde b } \nonumber \\ x_{72 } & = & \frac{1}{8\pi^2 } \frac{g_{a12}^2}{g_{abb}^2 } t d_{123}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2},m_{\tilde b_1 } ) \sin^2\theta_{\tilde b } \nonumber \\ & + & \frac{1}{8\pi^2 } \frac{g_{a12}^2}{g_{abb}^2 } t d_{123}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1},m_{\tilde b_2 } ) \cos^2\theta_{\tilde b } \nonumber \\ x_{73 } & = & -\frac{1}{8\pi^2 } g^2_{a12 } m_{\tilde g } d_{0}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2},m_{\tilde b_1 } ) \sin\theta_{\tilde b}\cos\theta_{\tilde b}\nonumber \\ & + & \frac{1}{8\pi^2 } g^2_{a12 } m_{\tilde g } d_{0}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1},m_{\tilde b_2 } ) \sin\theta_{\tilde b}\cos\theta_{\tilde b}\nonumber \\ x_{74 } & = & -\frac{1}{8\pi^2 } g^2_{a12 } m_{\tilde g } d_{0}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2},m_{\tilde b_1 } ) \sin\theta_{\tilde b}\cos\theta_{\tilde b}\nonumber \\ & + & \frac{1}{8\pi^2 } g^2_{a12 } m_{\tilde g } d_{0}(-p_1,p_3,p_4 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1},m_{\tilde b_2 } ) \sin\theta_{\tilde b}\cos\theta_{\tilde b}\nonumber \\ \nonumber \\ x_{8i } & = & x_{7i } ( t \leftrightarrow u , p_3 \leftrightarrow p_4 ) \quad i = 1,2,3,4 \nonumber \\ \nonumber \\
x_{93 } & = & -\frac{1}{8\pi^2 } g_{h11}g_{haa}\frac { m_{\tilde g}}{(s - m_h^2 ) + i m_h \gamma_h } c_{0}(-p_1,p_1+p_2 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_1 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b}\nonumber \\ & - & \frac{1}{8\pi^2 } g_{h12}g_{haa}\frac { m_{\tilde g}}{(s - m_h^2 ) + i m_h
\gamma_h } c_{0}(-p_1,p_1+p_2 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2 } ) \cos^2\theta_{\tilde b}\nonumber \\ & + & \frac{1}{8\pi^2 } g_{h12}g_{haa}\frac { m_{\tilde g}}{(s - m_h^2 ) + i m_h \gamma_h } c_{0}(-p_1,p_1+p_2 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1 } ) \sin^2\theta_{\tilde b}\nonumber \\ & + & \frac{1}{8\pi^2 } g_{h22}g_{haa}\frac { m_{\tilde g}}{(s - m_h^2 ) + i m_h \gamma_h } c_{0}(-p_1,p_1+p_2 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_2 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b}\nonumber \\ x_{94 } & = & -\frac{1}{8\pi^2 } g_{h11}g_{haa}\frac { m_{\tilde g}}{(s - m_h^2 ) + i m_h
\gamma_h } c_{0}(-p_1,p_1+p_2 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_1 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b}\nonumber \\ & + & \frac{1}{8\pi^2 } g_{h12}g_{haa}\frac { m_{\tilde g}}{(s - m_h^2 ) + i m_h \gamma_h } c_{0}(-p_1,p_1+p_2 , m_{\tilde g } , m_{\tilde b_1},m_{\tilde b_2 } ) \sin^2\theta_{\tilde b } \nonumber \\ & - & \frac{1}{8\pi^2 } g_{h12}g_{haa}\frac { m_{\tilde g}}{(s - m_h^2 ) + i m_h \gamma_h } c_{0}(-p_1,p_1+p_2 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_1 } ) \cos^2\theta_{\tilde b } \nonumber \\ & + & \frac{1}{8\pi^2 } g_{h22}g_{haa}\frac { m_{\tilde g}}{(s - m_h^2 ) + i m_h \gamma_h } c_{0}(-p_1,p_1+p_2 , m_{\tilde g } , m_{\tilde b_2},m_{\tilde b_2 } ) \sin\theta_{\tilde b } \cos\theta_{\tilde b}\nonumber \\ \nonumber \\ x_{10i } & = & x_{9i } ( g_{hjk } \leftrightarrow g_{hjk } , g_{haa } \leftrightarrow g_{haa } ) \quad i=3 , 4 ; \quad j , k = 1 , 2 \nonumber \\ \nonumber \\\end{aligned}\ ] ]
@xmath151 @xmath152 & @xmath153 & @xmath154 @xmath161 + \frac{g m_b^2}{m_w \cos\beta } \sin\alpha $ ] & & @xmath162 $ ] @xmath163 & @xmath164 + \frac{g m_b^2}{m_w \cos\beta } \sin\alpha $ ] & & @xmath165 $ ] @xmath166 & @xmath167 $ ] & & @xmath168 @xmath169 & @xmath170 - \frac{g m_b^2}{m_w \cos\beta } \cos\alpha $ ] & & @xmath171 $ ] @xmath172 & @xmath173 & & @xmath174 $ ] @xmath175 & @xmath176 & & @xmath177 @xmath178 & @xmath179 @xmath180 & @xmath181 20 h. p. nilles , phys
. rept .
* 110 * , 1 ( 1984 ) ; j. f. gunion , h. e. haber , g. l. kane and s. dawson , _ the higgs hunter s guide _ ( addison - wesley , menlo park , 1990 ) .
d. a. dicus , c. kao and s. s. d. willenbrock , phys .
b * 203 * , 457 ( 1988 ) .
e. w. n. glover and j. j. van der bij , nucl .
b * 309 * , 282 ( 1988 ) .
t. plehn , m. spira and p. m. zerwas , nucl .
b * 479 * , 46 ( 1996 ) [ erratum - ibid .
b * 531 * , 655 ( 1998 ) ] [ arxiv : hep - ph/9603205 ] .
s. dawson , s. dittmaier and m. spira , phys .
d * 58 * , 115012 ( 1998 ) [ arxiv : hep - ph/9805244 ] .
a. belyaev , m. drees , o. j. p. eboli , j. k. mizukoshi and s. f. novaes , phys .
d * 60 * , 075008 ( 1999 ) [ arxiv : hep - ph/9905266 ] .
a. a. barrientos bendezu and b. a. kniehl , phys .
d * 64 * , 035006 ( 2001 ) [ arxiv : hep - ph/0103018 ] .
t. binoth , s. karg , n. kauer and r. ruckl , phys .
d * 74 * , 113008 ( 2006 ) [ arxiv : hep - ph/0608057 ] .
s. dawson , c. kao , y. wang and p. williams , phys .
d * 75 * , 013007 ( 2007 ) [ arxiv : hep - ph/0610284 ] . l. g. jin , c. s. li , q. li , j. j. liu and r. j. oakes , phys . rev .
d * 71 * , 095004 ( 2005 ) [ arxiv : hep - ph/0501279 ] .
f. boudjema and e. chopin , z. phys .
c * 73 * , 85 ( 1996 ) [ arxiv : hep - ph/9507396 ] .
a. djouadi , w. kilian , m. muhlleitner and p. m. zerwas , eur .
j. c * 10 * , 45 ( 1999 ) [ arxiv : hep - ph/9904287 ] .
m. muhlleitner and m. spira , phys .
d * 68 * , 117701 ( 2003 ) .
u. baur , t. plehn and d. l. rainwater , phys .
d * 69 * , 053004 ( 2004 ) [ arxiv : hep - ph/0310056 ] .
m. moretti , s. moretti , f. piccinini , r. pittau and a. d. polosa , jhep * 0502 * , 024 ( 2005 ) [ arxiv : hep - ph/0410334 ] .
maltoni , f. and sullivan , z. and willenbrock , s. , phys .
d * 67 * , 093005 ( 2003 ) [ arxiv : hep - ph/0301033 ] .
j. campbell _ et al .
_ , [ arxiv : hep - ph/0405302 ] . s. dawson , c. b. jackson , l. reina and d. wackeroth , mod . phys
. lett . a * 21 * , 89 ( 2006 ) [ arxiv : hep - ph/0508293 ] .
s. dawson , c. b. jackson , l. reina and d. wackeroth , int .
j. mod .
a * 20 * , 3353 ( 2005 ) [ arxiv : hep - ph/0409345 ] .
s. dawson , c. b. jackson , l. reina and d. wackeroth , phys .
d * 69 * , 074027 ( 2004 ) [ arxiv : hep - ph/0311067 ] .
s. dawson , c. b. jackson , l. reina and d. wackeroth , phys .
lett . * 94 * , 031802 ( 2005 ) [ arxiv : hep - ph/0408077 ] .
s. dittmaier , m. kramer and m. spira , phys .
d * 70 * , 074010 ( 2004 ) [ arxiv : hep - ph/0309204 ] .
s. dawson and c.
b. jackson , arxiv:0709.4519 [ hep - ph ] .
r. m. barnett , h. e. haber and d. e. soper , nucl .
b * 306 * , 697 ( 1988 ) .
t. stelzer , z. sullivan and s. willenbrock , phys .
d * 58 * , 094021 ( 1998 ) [ arxiv : hep - ph/9807340 ] .
d. dicus , t. stelzer , z. sullivan and s. willenbrock , phys .
d * 59 * , 094016 ( 1999 ) [ arxiv : hep - ph/9811492 ] .
g. t hooft and m. j. g. veltman , nucl .
b * 44 * , 189 ( 1972 ) .
g. t hooft and m. j. g. veltman , nucl .
b * 153 * , 365 ( 1979 ) .
g. j. van oldenborgh and j. a. m. vermaseren , z. phys .
c * 46 * , 425 ( 1990 ) .
g. j. van oldenborgh , comput .
commun .
* 66 * , 1 ( 1991 ) .
m. a. aivazis , j. c. collins , f. i. olness and w. k. tung , phys .
d * 50 * , 3102 ( 1994 ) [ arxiv : hep - ph/9312319 ] .
j. c. collins , phys .
d * 58 * , 094002 ( 1998 ) [ arxiv : hep - ph/9806259 ] .
m. kramer , f. i. olness and d. e. soper , phys .
d * 62 * , 096007 ( 2000 ) [ arxiv : hep - ph/0003035 ] .
e. braaten and j. p. leveille , phys .
d * 22 * , 715 ( 1980 ) .
w. beenakker , a. denner , w. hollik , r. mertig , t. sack and d. wackeroth , nucl .
b * 411 * , 343 ( 1994 ) .
d. m. pierce , arxiv : hep - ph/9805497 .
p. hafliger and m. spira , nucl
. phys .
b * 719 * , 35 ( 2005 ) [ arxiv : hep - ph/0501164 ] .
s. berge , w. hollik , w. m. mosle and d. wackeroth , susy qcd one - loop effects in ( un)polarized top - pair production at hadron phys . rev .
d * 76 * , 034016 ( 2007 ) [ arxiv : hep - ph/0703016 ] . l. j. hall , r. rattazzi and u. sarid , phys . rev .
d * 50 * , 7048 ( 1994 ) [ arxiv : hep - ph/9306309 ] .
m. carena , d. garcia , u. nierste and c. e. m. wagner , nucl .
b * 577 * , 88 ( 2000 ) [ arxiv : hep - ph/9912516 ] .
j. guasch , p. hafliger and m. spira , phys .
d * 68 * , 115001 ( 2003 ) [ arxiv : hep - ph/0305101 ] .
m. s. carena , a. menon , r. noriega - papaqui , a. szynkman and c. e. m. wagner , phys .
d * 74 * , 015009 ( 2006 ) [ arxiv : hep - ph/0603106 ] .
j. pumplin , d. r. stump , j. huston , h. l. lai , p. nadolsky and w. k. tung , jhep * 0207 * , 012 ( 2002 ) [ arxiv : hep - ph/0201195 ] .
w. j. marciano , phys .
d * 29 * , 580 ( 1984 ) .
p. nason , s. dawson and r. k. ellis , nucl .
b * 303 * , 607 ( 1988 ) .
j. a. m. vermaseren , s. a. larin and t. van ritbergen , phys .
b * 405 * , 327 ( 1997 ) [ arxiv : hep - ph/9703284 ] .
h. e. haber and g. l. kane , phys .
rept . * 117 * , 75 ( 1985 ) .
m. drees , r. m.godbole and p. roy , _ theory and phenomenology of sparticles : an account of four - diemnsional n=1 supersymmetry in high energy physics _
( world scientific ) h. baer , m. bisset , c. kao and x. tata , phys . rev .
d * 46 * , 1067 ( 1992 ) .
h. e. haber , m. j. herrero , h. e. logan , s. penaranda , s. rigolin and d. temes , phys .
d * 63 * , 055004 ( 2001 ) [ arxiv : hep - ph/0007006 ] .
|
we present a complete next - to - leading order ( nlo ) calculation for the total cross section for inclusive higgs pair production via bottom - quark fusion at the cern large hadron collider ( lhc ) in the minimal supersymmetric standard model ( mssm ) and the minimal supergravity model ( msugra ) .
we emphasize the contributions of squark and gluino loops ( sqcd ) and the decoupling properties of our results for heavy squark and gluino masses .
the enhanced couplings of the @xmath0 quark to the higgs bosons in supersymmetric models with large @xmath1 yield large nlo sqcd corrections in some regions of parameter space .
|
weakly interacting massive particles ( wimps ) are among the leading candidates for the dark matter ( dm ) component in the universe .
the framework is elegant and simple : stable wimps can be embedded in most extensions to the standard model of particle physics . in thermal equilibrium in the early universe ,
they decouple from the primordial bath in the non - relativistic regime .
their relic abundance scales approximatively with the inverse of their total pair annihilation rate into lighter particles : the weak - interaction coupling ensures that , within the standard cosmological scenario , such relic density is of the order of the mean density of dm in the universe today , as determined in cosmological observations ( for comprehensive reviews on wimp dm candidates and their detection , see , e.g. , @xcite ) in principle , one of the routes to test the hypothesis of wimp dm stems from the bases of the framework themselves .
supposing that wimps are indeed the building blocks of all structures in the universe , there is a ( small but finite ) probability that wimps in dm halos , including the halo of the milky way , annihilate in pairs into detectable species .
indirect detection has mainly been focused on the search for a wimp - induced component in the local antiproton , positron , and antideuteron cosmic - ray fluxes and for an excess in the high - energy gamma - ray galactic or extra - galactic flux ( relevant constraints on the wimp parameter space have been derived from such analyses ; for recent results , see , e.g. , @xcite ) .
since the gamma - ray signal scales with the square of the wimp density along the line of sight , the galactic center ( gc ) has been often indicated as the prime target . in any self - consistent model for the distribution of dm in galactic halos , the dm density is found to be maximal at the center of the system ; numerical n - body simulations of hierarchical clustering in cold dark matter cosmologies find configurations with very large overdensities , consistent with singular density profiles @xcite ( notice , however , that the simulations lack resolution to map the distribution of dm on the very small scales which are relevant for wimp signals ; at the same time , there is some tension between the simulation results and the dm profiles as derived from dynamical observations for certain classes of galaxies , see , e.g. @xcite ) .
the galactic center is an extraordinary site from several different points of view .
dynamical observations point to the presence of a supermassive black hole @xcite , with mass @xmath1 , located very close to the dynamical center of the galaxy , and most likely associated to the compact radio source labeled sgr a@xmath0 .
infrared and x - ray counterparts have been identified for sgr a@xmath0 ; gev and tev emissions in the direction of the gc have been detected as well , with the first data with high statistics and fair angular resolution which have been obtained with the hess air cherenkov telescope @xcite .
sgr a@xmath0 is an unusual source , certainly very different from typical galactic or extragalactic compact sources associated to black holes .
most notably , under our perspective , it has a very low luminosity over the whole spectrum , at a level at which it is plausible that a wimp - induced component may be relevant .
the emission associated to wimp annihilations is expected to extend from the radio band up to gamma - ray frequencies .
the peak of the gamma - ray luminosity stands at the energy corresponding to a fraction ( say one - third to one - twentieth ) of the wimp mass , which is in turn in the few ( tens of ) gev few tev range ; it is mostly associated to the chain of decays and/or hadronization processes initiated by two - body final state particles from wimp pair annihilations , leading to the production of neutral pions and their subsequent decays into two photons . in an analogous chain , with analogous efficiency ,
high - energy electrons and positrons are produced by charged pions . emitted in a region with large magnetic fields , as typical for
accretion flows around supermassive black holes , they give rise to synchrotron emission covering radio frequencies up to , possibly , the x - ray band .
numerous analyses have been dedicated to the study of the gc as a wimp gamma - ray source , a list of recent references includes , e.g. , @xcite .
a prediction for the synchrotron emission has been discussed in refs .
@xcite , and refined on several aspects in ref .
@xcite ; a comparison with x - ray data motivated by a class of heavy wimp dm candidates is presented in ref @xcite .
we consider here the topic within a self - consistent multi
wavelength approach . referring to a generic wimp dm scenario
, we discuss spectral and angular features , and sketch the correlations among signals in the different energy bands .
we illustrate which are the critical assumptions in deriving such conclusions , analyze them in the context of the currently available datasets , and make projections for the testability of the framework in the future .
the procedure we follow is , to some extent , analogous to the one adopted in ref .
@xcite for galaxy clusters , and in ref .
@xcite for dwarf satellites ; the list of relevant references include also @xcite .
the paper is organized as follows : in section [ sec : data ] we briefly summarize currently available observations on the gc . in section [ sec : source ]
we describe the dm source and the related mechanisms of photon production . in section
[ sec : scalings ] we compute the approximate scalings of the multi wavelength spectrum , while in section [ sec : fulltreat ] we perform the full treatment for some benchmark models .
then in section [ sec : result ] we compare the dm induced signal with the present limits and with the projected constraints of forthcoming experiments .
finally , section [ sec : conclusion ] concludes .
the radio to sub - mm emission from sgr a@xmath0 is characterized by a very hard spectrum : the luminosity above @xmath2 ghz scales approximately as @xmath3 with @xmath4 and cut - off at about @xmath5 ghz ( a compilation of available data and a full list of references is given , e.g .. in ref .
@xcite ) . we will show that such features do not seem to be compatible with the synchrotron emission induced by wimp annihilations , not even with the observed flux reshaped by synchrotron self - absorption .
in general , softer spectra are obtained , and the comparison with observations is useful to infer limits on the wimp parameter space .
the tightest bound follows from the measurement at the lowest frequency , i.e. the upper bound on the flux density @xmath6 jy at the frequency @xmath7 mhz , obtained with an interferometer with 4.3 arcsec angular acceptance at jodrell bank @xcite .
although variations are clearly seen in the radio flux density at different epochs , clean patterns of temporal dependencies have not been identified , see , e.g. , @xcite ; the data plotted in fig . [
fig : data ] are not time - averaged . at any given frequency , we show , among the available measurements , the one corresponding to the epoch of lowest luminosity .
the angular size of the source depends on the frequency of observation . at 1 ghz ,
a frequency at which it is expected that scattering in the interstellar medium would wash out the true structure of sgr a@xmath0 , it is of the order of 1.5 arcsec @xcite . at higher frequencies ,
the size shrinks proportionally to @xmath8 up to the measured value of 0.2 mas ( about 1 au in physical size ) at about 86 ghz @xcite , possibly at the level of the intrinsic size of the source .
we will show that , at radio frequencies , the angular size of a wimp - induced component is expected to be much larger than these apparent or intrinsic angular dimensions ; we find sizes for which it is actually interesting to compare with wide field images of the gc region rather than sgr a@xmath0 alone . among the available surveys
, we will refer again to those at the lowest frequency , namely at 90 cm .
an atlas of the diffuse radio emission in the milky way was presented in @xcite .
the evidence for a gc diffuse non - thermal source was enlighted in @xcite .
both of the maps have an angular resolution @xmath9 , thus hiding the spatial structure of the diffuse emission in the innermost region .
we will consider an image of the gc region constructed from vla data , covering an area of @xmath10 degrees and with angular resolution of 43 arcsec @xcite .
the near - infrared and x - ray emissions from sgr a@xmath0 are characterized by a large variability ( on different timescales in the two cases ) : quiescent values for the luminosity are plotted in fig .
[ fig : data ] .
the quiescent flux in the near - infrared has been recently detected with the vlt @xcite as a point source with a position coincident with the supermassive black hole within an accuracy of 10 - 20 mas , limited by faintness and by the proximity of one of the stars orbiting the black hole @xcite . launched in 1999 ,
nasa s chandra x ray observatory is at present the most powerful x ray detector , covering the energy range 0.1 kev10 kev with an angular resolution of 0.5 arcsec . during its observations
, it has clearly discovered an x ray source consistent with the position of sgr a@xmath0 @xcite , whose quiescent emission is well fitted by an absorbed thermal bremsstrahlung plus a gaussian - line , plotted in fig . [
fig : data ] .
the spatial dimension of the x - ray source is @xmath11 arcsec .
the process involving wimp annihilations is expected to be steady , i.e. it can not reproduce any time variability pattern .
we will show that a x ray flux at the quiescent level detected by chandra can be obtained in the case of large wimp densities and large magnetic field ; moreover the source is predicted essentially as point like , rather than the extended source seen by the chandra detector .
we will use sgr a@xmath0 infrared and x ray data to set constraints on wimp models .
chandra detected also a diffuse emission in several regions within the inner 20 pc of the galaxy .
the reconstructed image covers a field of view of @xmath12 around sgr a@xmath0 @xcite .
this diffuse emission could be consistently modeled as originating from a two temperature diffuse plasma .
the soft component ( @xmath13 kev ) could be explained invoking different astrophysical mechanisms , while the origin of the hard component ( @xmath14 kev ) , spatially uniform , is not clearly understood . in principle
it could be explained in terms of inverse compton scattering on cosmic microwave background ( cmb ) induced by wimp annihilations ; however the detection of several emission lines and the inconsistency with limits at other frequencies make this hypothesis unplausible .
we come finally to gamma - ray observations .
the egret team has reported the observation of a gc source in the energy range 100 mev20 gev @xcite . as it can be seen in fig .
[ fig : data ] , the luminosity of such a source exceeds by about one order of magnitude the luminosity of sgr a@xmath0 at any other frequency .
the angular resolution of egret was rather poor , about 1 degree at 1 gev , encompassing a large portion of the gc and not allowing for a clean identification of the emitter . in ref .
@xcite , the authors argue that the improvement of the instrument angular resolution at multi - gev energies should be taken into account in the data analysis , and conclude that it is actually possible to exclude the identification of the egret source with sgr a@xmath0 ; in the same paper it is suggested that the comparison to set constraints on wimp models should be with the diffuse background measured by egret in the gc region , rather than with the egret gc source .
the detection of tev gamma - ray radiation from the gc has been reported by hess @xcite .
such a measurement has been confirmed , with a consistent spectrum , by magic @xcite and supersedes previous results by cangaroo @xcite and whipple @xcite , whose significantly different spectra is likely due to a miscalibration of the detector and poorer statistics rather than variability of the source .
hess has discovered a point source , whose position is coincident with sgr a@xmath0 within 7.3 arcsec @xmath15 8.7 arcsec ( stat . )
@xmath15 8.5 arcsec ( syst . )
@xcite , excluding the identification with the nearby supernova remnant sgr a east , but not with other candidates , such as a pulsar wind nebula recently discovered by chandra @xcite which is only 8.7 arcsec away from sgr a@xmath0 .
the luminosity spectrum of the hess point source is shown in fig . [
fig : data ] ; it is a rather features - less flux , @xmath16 with spectral index @xmath17 , extending from 160 gev up to above 20 tev . even on the basis of the spectral characteristics only , without any consistency checks at other wavelength , it has been shown that it is rather implausible that such a source is due to wimp annihilations only @xcite .
hess has also reported the detection of a diffuse gamma - ray emission along the central 300 pc of the gc ridge , within about 0.8 degree in longitude and 0.3 degree in latitude with respect to the gc .
we will consider the central source and the diffuse emission as maximal background level to understand the potential for a discovery of a wimp component with upcoming gamma - ray telescopes . .
the radio to x ray emissions are shown in the quiescent state or at the epoch of lowest luminosity among available observations .
the plotted @xmath18ray sources have positions compatible with sgr a@xmath0 ; however , due to a poor angular resolution , egret can not clearly identify the source and perhaps neither the hess telescope .
see the text for details about the observations in each band.,width=10 ]
the dm wimp source scales with the number density of wimp pairs locally in space , i.e. assuming a smooth ( i.e. without substructures ) , spherically symmetric , and static dark matter distribution , with @xmath19 , with @xmath20 being the halo mass density profile at the radius @xmath21 , and @xmath22 the mass of the dark matter particle .
emitted stable species are nearly monochromatic if they are direct products of the annihilation ( since the annihilating particles are essentially at rest ) ; they have much broader spectra if they are generated in cascades with decays and/or hadronization processes of unstable two - body final states . for a given species @xmath23 ,
the source function takes the form : q_i(e , r)=(v ) ( e ) , [ eqq ] where @xmath24 is the annihilation rate at zero temperature , and @xmath25 is the number of particles @xmath23 emitted per annihilation in the energy interval @xmath26 , obtained by weighting spectra for single annihilation channels over the corresponding branching ratio .
the species which are relevant in a multi wavelength analysis are photons , as well as electrons and positrons which act as sources for radiative processes . for most wimp models , branching ratios for monochromatic emission in these channels
are subdominant ; in our analysis , we will concentrate on the components with continuum spectra .
prompt emission of photons proceeds mainly through the production and decay of neutral pions , hence the energy of these photons is in the gamma - ray band . for reference and to make transparent the connection with the notation introduced below for radiative processes
, the @xmath18ray emissivity can be written as : j_(e , r)=q_(e , r)e . in the energy range of interest for this analysis absorption
is negligible , and fluxes or intensities can be straightforwardly derived summing contributions along the line of sight .
e.g. , the differential @xmath18ray flux is : _ ( e,)=_l.o.s.ds [ eq : gammafl ] where the coordinate @xmath27 runs along the line of sight and @xmath28 is the angular off set from the galactic center . momentum .
synchrotron losses are shown for two reference values for the magnetic field : @xmath29 .
radiative losses associated to bremsstrahlung , ionization and coulomb scattering are shown at the gc ( lower curves ) and at a distance of 100 pc from the gc ( upper curves ) . _
right panel : _ distance @xmath30 travelled by an electron with an injection energy of 1 gev before losing most of its energy ; three different guesses for the diffusion coefficient are shown , in the case of equipartition and reconnection magnetic field , see fig .
[ fig : bfield]a ( same line styles).,width=294 ] momentum .
synchrotron losses are shown for two reference values for the magnetic field : @xmath29 .
radiative losses associated to bremsstrahlung , ionization and coulomb scattering are shown at the gc ( lower curves ) and at a distance of 100 pc from the gc ( upper curves ) . _
right panel : _ distance @xmath30 travelled by an electron with an injection energy of 1 gev before losing most of its energy ; three different guesses for the diffusion coefficient are shown , in the case of equipartition and reconnection magnetic field , see fig .
[ fig : bfield]a ( same line styles).,width=294 ] the emission through radiative losses involves charged particles , mainly electrons and positrons . produced in wimp
pair annihilations , they propagate , losing and/or gaining energy . to describe this process , we consider the transport equation , in the limit of spherical symmetry , and for a stationary solution
( see for example @xcite ; diffusive reacceleration is neglected ) : - + v -(r^2v)p+(p p^2 f)= q ( r , p ) [ eqtransp ] where @xmath31 is the @xmath32 distribution function at the equilibrium , at a given radius @xmath21 and in terms of the momentum @xmath33 , related to the number density in the energy interval @xmath26 by : @xmath34 ; analogously , for the wimp source function of electrons or positrons , we have @xmath35 . the first term on the left - hand side ( l.h.s .
) describes the spatial diffusion , with @xmath36 being the diffusion coefficient .
the second and third terms model an advective ( convective ) transport with an inflow ( outflow ) of the electrons and positrons toward ( away from ) the center of the system , being @xmath37 the flow velocity of the medium .
finally , the last term on the l.h.s . describes the energy loss of due to radiative processes ; @xmath38 is the sum of the rates of momentum loss associated to the radiative process @xmath23 .
we apply eq .
[ eqtransp ] to the gc .
the radiative losses affecting the @xmath32 propagation are synchrotron emission , inverse compton scattering on cmb and starlight , bremsstrahlung , ionization , and coulomb scattering .
we model the galactic medium as composed by molecular ( h@xmath39 ) , atomic ( hi ) , and ionized ( hii ) gases .
the density profiles are extracted from the description of the central molecular zone in @xcite , approximating their results under the assumptions of spherical symmetry .
the synchrotron loss rate is spatially dependent , scaling with the square of the local value of the magnetic field .
we plot in fig .
[ fig : enloss]a the time
scale for the energy loss associated to each radiative process , defined as @xmath40 .
we show the synchrotron emission for two reference values of the magnetic field , while the two curves for bremsstrahlung , ionization and coulomb scattering refer to the losses at the gc and at a distance of 100 pc from the gc .
we plot one curve for inverse compton scattering ; the time
scale is inversely proportional to the energy density of the background radiation ; at the gc the energy density of the starlight component is considerably larger ( @xmath41 ev @xmath42 @xcite ) with respect to the cmb ( @xmath43 ev @xmath42 ) .
we sum the two components assuming a starlight energy density constant over the whole gc region .
the radial profile of the magnetic field is indeed an important ingredient in our analysis .
based on observations of nonthermal radio filaments , polarization of thermal dust emission , and synchrotron radiation from cosmic rays , the canonical picture of the galactic center magnetosphere ( for a review , see @xcite ) describes the magnetic field with a dipolar geometry on large scale and as a pervasive field with strength of a mg throughout the central molecular zone ( few hundred of pc ) .
the recent discovery of a diffuse source of nonthermal synchrotron emission @xcite suggests , on the other hand , a mean magnetic field of order 10 @xmath44 g on scales @xmath45 few pc , unless reacceleration processes are invoked .
it s important to note that such analyses constrain the mean magnetic field on scales @xmath45 pc and do not exclude strong magnetic field in the innermost region .
following again @xcite , we consider a magnetic field for the gc region satisfying the equipartition condition , namely , with the magnetic energy completely balancing the kinetic pressure : b(r)=3.910 ^ 4 ( ) ^5/4 .
[ eqbfield ] from a conservative point of view , this could be regarded as the maximal allowed magnetic field ; we discuss this case together with two further possibilities : we follow @xcite and consider the case for a reduced magnetic field due to magnetic field line reconnection in turbulent plasma @xcite ; as a toy model of an extreme case at the other hand with respect to the equipartition assumption , we allow also for a magnetic field which is constant within the accretion region . outside the accretion region , assuming spherical infall and flux conservation , the magnetic field scales as @xmath46 up to the large scale value @xmath47 g @xcite . the three different choices for the magnetic field radial profile
are shown in fig .
[ fig : bfield]a . note that for magnetic fields b @xmath48 g ( as is typical for the innermost region of the galaxy ) , the synchrotron losses dominate at all energies . for lower magnetic fields ,
i.e. at larger scales , inverse compton scattering ( and bremsstrahlung ) becomes relevant in the ultra
relativistic regime , while ionization starts to dominate in the non relativistic limit . in order to estimate the relevance of spatial diffusion
, we compare in fig .
[ fig : enloss]b the physical scale @xmath21 with the distance diffused by electrons before losing most of their energy , @xmath49 . in the quasilinear approximation of turbulent diffusion
, the form of the diffusion coefficient @xmath50 can be expressed as @xmath51 , where @xmath52 is the gyroradius of the electron , @xmath53 is the electron velocity , and @xmath54 is the random component of the magnetic field at the resonant wavelength @xmath55 . on large scale ( i.e. , larger than about 100 pc )
cosmic - ray data seem to indicate that the diffusion coefficient takes the form : @xmath56 with @xmath57 and @xmath58 @xcite ; in the innermost region , the picture is much more uncertain .
indirect constraints are derived in the models of @xcite and @xcite , when addressing the origin of the @xmath18-ray source detected by hess at the gc ; in both analyses a significant reduction of the diffusion coefficient in the inner @xmath59 pc region is found . on the modelling side ,
the relevance of diffusion is strictly connected with unknown variables needed in the description of turbulence , namely , the amplitude of the random magnetic field and the scale and the turbulence spectrum . as an example , one can assume comparable strengths for the regular and the random components of the magnetic field , and a power law , @xmath60 for the turbulence spectrum . for bohm diffusion ( typical when the coherence length of the magnetic field is comparable or greater with respect to the gyroradius of electrons ) , @xmath61 and the coefficient reduces to @xmath62 ; as one can see in fig .
[ fig : enloss]b , in this case the effect of diffusion can be safely neglected . assuming a turbulent regime ( in a homogeneous medium ) with a scale of turbulence @xmath63
, we find that , for @xmath64 ( `` kolmogorov '' , i.e. assuming a random flow of an incompressible fluid ) and @xmath65 ( `` kraichnan '' , which is more likely than the kolmogorov spectrum in the case of the strong large - scale magnetic field ) , the diffusion can be relevant from the sub - pc scale in the first case , and it is marginally relevant around the pc scale in the second case . note the main ingredient here is the very large magnetic field considered in the innermost region of the galaxy . in our sample models , we find then that diffusion is either negligible over the whole central region or that it might be relevant only in its outer part , where , however , the dm source is expected to be less strong and have a less steep gradient than close to the central bh ( see the discussion below ) .
therefore , we can foresee negligible to very mild effects from diffusion and , in what follows , for sake of simplicity and to make the discussion clearer , we will disregard the diffusion term .
we describe the accretion flow of gas onto the black hole in the innermost part of the galaxy following @xcite , namely , we assume that the supersonic wind entering in the bh gravitational potential forms a bow shock dissipating kinetic energy and subsequently falls radially onto the bh .
we take a spherical accretion and a steady flow , and estimate the region of the accretion as @xmath66 , where @xmath67 km s@xmath68 is the galactic wind velocity and thus @xmath69 pc @xcite .
the radial infall velocity of the gas is v(r)=-c [ eqgasvel ] a particle propagating in such accretion flow gains momentum since it feels an adiabatic compression in the bh direction .
the galactic center lobe is a radio continuum emission spanning the central degree of the galaxy with a bipolar structure .
recent mid - infrared observations @xcite suggest the idea that the emission associated to the gc lobe is a sign of a gc outflow , in particular , a starburst outflow .
the associated large - scale bipolar wind could affect the transport equation eq .
[ eqtransp ] , convecting electrons and positrons . assuming
a velocity @xmath70 km / s @xcite , this effect is negligible in the innermost region , while it can be relevant on larger scales .
on the other hand , although the model of @xcite is probably the most intriguing , one can resort to other mechanisms explaining the origin of the galactic center lobe ( for a recent review , see , e.g. , @xcite ) . in the following we choose to neglect the effect of such a possible wind . the solution of eq .
[ eqtransp ] provides the @xmath71 number density @xmath72 in the stationary limit . for a radiative process @xmath23 , with associated power @xmath73 ,
the photon emissivity is given by folding @xmath72 with the power @xcite : j_i(,r)=2^m__m_ede p_i(r , e , ) n_e(r , e ) , [ eqjsynch ] where the factor @xmath74 takes into account electrons and positrons ( in wimp annihilations , as well as during propagation , there is perfect symmetry between particles and antiparticles ) .
for any given emission mechanism , the associated luminosity at frequency @xmath75 is l_i()=d^3rj_i(,r ) , [ eqlumin ] while the intensity measured by a detector can be estimated as s_i(,,_d ) = d ( -)_l.o.s.di_i(,s , ) .
[ eqisynch ] here @xmath28 labels the direction of observation , i.e. the angular off set with respect to the gc and we are performing an angular integral assuming a circular gaussian resolution of width @xmath76 for the detector .
@xmath77 is the differential of the intensity of radiation @xmath78 : within the increment @xmath79 along a line of sight , there is a gain in intensity @xmath80 , while a decrease @xmath81 could be due to absorption .
@xmath78 follows from the solution of the differential equation : = -(,s,)i_i(,s,)+ [ eqdiffisynch ] where @xmath82 is the angular off - set from the gc of the line of sight along which @xmath78 is calculated , as selected by @xmath28 and the angular variables of integration @xmath83 and @xmath84 . if absorption is negligible , the second integrand of eq .
[ eqisynch ] reduces to @xmath85 . at low and intermediate frequencies ,
i.e. in the radio band up to ( possibly ) the x - ray band , the dm signal is mostly due to synchrotron radiation .
the power for synchrotron emission takes the form @xcite : p_syn ( r , e,)= b(r ) f(/_c ) , where @xmath86 is the electron mass , the critical synchrotron frequency is defined as @xmath87 , and @xmath88 is the standard function setting the spectral behavior of synchrotron radiation . at radio frequencies ,
there are configurations for the parameters in the model for which synchrotron self - absorption is a relevant effect @xcite ; we include it in our analysis implementing the standard form for @xmath89 in this case . in
@xcite it was shown that , on the other hand , we can safely neglect synchrotron self - compton effects .
for the signal in the uv and soft x band , we need to take into account the photoelectric effect on the interstellar dust ; this is an effect taking place mostly outside the region of emission hence we can model it a posteriori through an attenuation factor
. the emission through inverse compton ( ic ) scattering of the ultra relativistic electrons from wimp annihilations on cosmic microwave or starlight background photons , could be relevant as well , especially when targeting the gc region rather than the sgr a@xmath0 only .
this emission spans the x - band up to the ( soft ) @xmath18-ray band .
the inverse compton power is given by p_ic(r , e , ) = chd ( , r)(,,e ) [ eqpic ] where @xmath90 is the energy of the target photons , @xmath91 is their differential energy spectrum , and @xmath92 is the klein
nishina cross section .
finally , a very faint emission is expected in case of bremsstrahlung , ionization , and coulomb scattering ; we will not consider them in our analysis .
ray and @xmath32 spectra per annihilation for a @xmath93 tev wimp .
the three annihilation channels @xmath94 , @xmath95 , and @xmath96 are taken as references . _
right panel : _ multiplicity between the electron and photon yields @xmath97 for a @xmath93 tev wimp with the same annihilation modes as in the left panel.,width=294 ] ray and @xmath32 spectra per annihilation for a @xmath93 tev wimp .
the three annihilation channels @xmath94 , @xmath95 , and @xmath96 are taken as references . _
right panel : _ multiplicity between the electron and photon yields @xmath97 for a @xmath93 tev wimp with the same annihilation modes as in the left panel.,width=294 ] in this section we sketch in a simple and analytic form the scalings of the dark matter induced signal depending on various assumptions in the model .
. [ eqtransp ] does not admit in general an analytic solution .
however , when the radiative loss term dominates ( and thus the first three terms are negligible ) , one finds simply : n_e(r , e)=^m__ede q_e(e , r ) [ eqdnde ] where @xmath98 comes from @xmath99 in eq .
[ eqtransp ] mapping momentum into energy .
we have already stressed that synchrotron processes are the main effect for energy losses and radiative emissivity .
we can focus , for the moment , on this mechanism , and write the energy loss rate as @xmath100 , and the induced synchrotron luminosity as l_^syn=4dr r^2 ( r)^2 ^m__e y_e(e ) = dr r^2 ( r)^2 e_p y_e(e_p ) + [ eq : nulnus ] where we have defined @xmath101 , and in the last step we have implemented the monochromatic approximation for the synchrotron power , i.e. assumed @xmath102 @xcite . in the monochromatic approximation there is a one - to - one correspondence between the energy of the radiating electron ( peak energy in the power ) and the frequency of the emitted photon , that is @xmath103 with @xmath104 , or , introducing values for numerical constants , the peak energy in gev is @xmath105 , with @xmath106 the frequency in ghz and @xmath107 the magnetic field in mg .
analogously , the induced @xmath18ray luminosity is l_^=2 dr r^2 ( r)^2e^2 .
[ eq : nulnug ] it is useful to make a few simple guesses on some of the quantities introduced above . along the line of @xcite , we assume the @xmath18ray spectrum per annihilation following the law : @xmath108 , with @xmath109 . it is also a fair assumption to approximate the integrated @xmath32 yield as a power law plus an exponential cutoff : @xmath110 . the differential yields of secondary photons and @xmath32 are plotted in fig . [
fig : yield]a , for three sample cases of two - body final states from wimp pair annihilations .
these plots are obtained linking to simulations of decay / hadronization performed with the pythia monte
carlo package @xcite and stored libraries contained in the darksusy package @xcite ; we will refer to such kind of simulations everywhere in the paper when making detailed estimates of wimp induced signals . as the simplest guess for radial dependence for the magnetic field and the dm profile , we consider the single power - law scalings , @xmath111 and @xmath112 .
[ eq : nulnus ] and [ eq : nulnug ] become : \ { l_^syn= & _ 0 ^ 2a^2 dr r^2 - 2()^(1-b ) expgev + l_^=&2a _ 0 ^ 2a^2 dr r^2 - 2expgev .
[ eq : nulnusg ] with @xmath113 the wimp mass in gev .
the right - hand - sides of eq .
[ eq : nulnusg ] show some differences . for the gamma - ray luminosity
, the energy cutoff follows simply from energy conservation and thus scales with the dark matter mass , except for a o(1 ) factor related to the annihilation mode . for synchrotron emission , at a fixed mass
, the frequency cutoff increases with the magnetic field , again except for a o(1 ) factor related to the annihilation channel .
away from the cutoff , the synchrotron emissivity tends to originate from a larger spatial region with respect to the @xmath18ray case , due to the additional positive power @xmath114 in the radial dependence . at fixed mass and frequency ,
if the magnetic field is large enough to avoid the frequency cutoff , the synchrotron signal is wider than the gamma - ray signal .
this is typically the case in the radio band and , to a much smaller extent , in the infrared band .
going to very high observed frequencies , however , the magnetic field ( or the energy of the radiating electron or positron ) needs to increase to exceedingly large values , which might be met only very close to the central bh ( or for extremely massive wimps and/or hard @xmath32 spectrum , as encoded in the factor @xmath115 of eq .
[ eq : nulnusg ] ) .
scalings of the required magnetic field , as a function of peak radiating energy , for a few values of the observed frequency are shown in fig .
[ fig : bfield]b : one can see that for the observed frequencies getting into the x - ray band ( say @xmath116 hz ) a very small radial interval is selected , corresponding to the largest allowed value for the magnetic field .
hence , in this case the synchrotron signal is actually expected to be originated in a very small region around the central bh , possibly much smaller compared to the gamma - ray flux .
we can now make a sketchy estimate to find which of the limits in the different bands in fig .
[ fig : data ] might be more constraining .
we write the ratio between synchrotron and gamma - ray luminosity in the form : r= = .
[ eq : nulfrac ] are the same as in the benchmark models.,width=10 ] in fig .
[ fig : yield]b we plot the relative multiplicity between photons and electrons for the three benchmark final states from wimp pair annihilations considered in fig .
[ fig : yield]a .
this illustrates the fact that , sufficiently far away from the energy cutoff and for a generic wimp annihilation channel ( except , of course , for the case of prompt emission of monochromatic gammas , and/or electrons / positrons we are not considering here ) , the photon and electron / positron yields are comparable and hence that it is difficult to avoid the correlation between the gamma and the synchrotron signals by selecting a specific wimp model . in eq .
[ eq : nulfrac ] this implies that the ratio @xmath117 is typically o(1 ) . the last term in eq .
[ eq : nulfrac ] does critically enter in boosting or suppressing the ratio of luminosities only in case the exponential cutoff ( or the upper limit in the radial integral ) is playing a role , i.e. at very large observational frequencies for synchrotron emission ( the x - ray band ) or for shallow density profiles . restricting to the case of singular halo profiles , and ,
e.g. the radio band , it is of order o(1 ) or o(0.1 ) . to see this more precisely , let us take @xmath95 as the annihilation channel , as an intermediate case between the soft quark spectra and the hard leptonic spectra .
we find that integrated @xmath32 yield , for masses in the range @xmath22 = 100 gev10 tev , can be fairly well approximated by @xmath118 ; the differential @xmath18spectrum was fitted in @xcite , finding @xmath119 . since @xmath120 , the rescaling factor @xmath121^{-(1-b)/2}$ ] , in the integral in the numerator of eq .
[ eq : nulfrac ] , varies at most between about 0.09 and 3 , hence we can assume as a sample value for the suppression expected from the ratio of integrals a factor of 0.5 . inserting all fit parameters in eq .
[ eq : nulfrac ] , we get : r_w ~910 ^ -2 ( ) ^0.2 ( ) ^0.15 ( ) ^0.5 [ eq : fracw ] we find hence that the radio and @xmath18ray luminosities are at a comparable level , while as it can be seen in fig .
[ fig : data ] , constraints in the @xmath18ray band are several orders of magnitude weaker than at radio wavelengths . although the luminosities of eqs .
[ eq : nulnus ] and [ eq : nulnug ] can not be directly compared with such experimental data , since they are integrated over the whole emission region , which can be significantly larger than the angular acceptance in the observations , and relevant effects such as advection and synchrotron self - absorption have been neglected , our approximate result in eq .
[ eq : fracw ] puts us on the track that the strongest constraints on the wimp parameter space should be related to synchrotron emission .
.benchmark models . [ cols="^,^,^,^,^,^",options="header " , ] the milky way is the galaxy we know in furthest detail , still the determination of its dm halo profile is not a simple task , in particular , for what concerns its inner region .
as is well known , there is some tension between n body simulation results suggesting the presence of a sharp enhancement in the central density , with dynamical observations of spiral galaxies finding configurations consistent with a density profile with a constant central core .
one of the uncertainties in implementing results from n body simulations regards the interplay between dark matter and the baryonic components of the galaxy ; in particular , the formation of the supermassive black hole ( smbh ) at the galactic center could have strongly modified the initial dm profile .
the adiabatic growth of a black hole at the center of a singular halo profile leads to the accretion of a very dense dm spike around it @xcite .
although this picture requires tuned initial conditions @xcite ( smbh seed very close to the center of the dark matter distribution and slow adiabatic growth ) , it is actually not excluded and , if the spike is formed , it can be completely destroyed only in a major merger event , unlikely in the recent past of the galaxy . the picture sketched in @xcite and
@xcite has been further refined in @xcite , where a time - evolution analysis of the cusp formation is performed , including the effects of self annihilations , scattering of dark matter particles by stars , and capture in the black hole .
the presence of relatively large overdensity in the galactic center region is an essential ingredient for a sizable wimp dark matter signal at any of the wavelengths we will consider in our treatment .
we follow the analysis in @xcite and focus our attention on two distributions obtained from the evolution of a navarro - frenk - white ( nfw ) profile @xcite : in the first ( hereafter labeled @xmath122 ) we include the formation of a density spike around the smbh only , while the second profile ( hereafter labeled @xmath123 ) is obtained by taking into account the deepening in the galactic potential well generated by the slow adiabatic formation of the stellar component in the inner galaxy , as well as that of the smbh . in this second case
the stellar component itself leads to a steepening of the halo profile from @xmath124 into @xmath125 @xcite ; such a steepening and ignoring a back
reaction on the dm profile tout
court stands as a limiting case among the series of results that have been obtained for the back reaction effect in the literature , starting from different assumptions and using either analytic treatments or numerical simulations , see e.g. @xcite , @xcite and @xcite . for both the @xmath122 and @xmath123 profiles the effect of self annihilation triggers the density in the innermost region , with the final shape being fixed by the value of the ratio @xmath126 @xcite .
there is therefore in general a non linear dependence of the wimp dm source , see eq .
[ eqq ] , on the cross section ( we have implemented such dependence in our analysis using scaling laws derived from either results given in @xcite or further sample cases kindly provided by the authors of @xcite ) .
our benchmark dm profiles are shown in fig . [ figprofiles ] , together , for comparison , with the nfw profile and the `` spiky '' profile obtained implementing the original procedure outlined in @xcite .
sample values for the wimp mass and the annihilation cross section are chosen here such that the multi
wavelength constraints are not violated ( verified a posteriori in section [ sec : result ] ) .
we do not focus our discussion on specific wimp models , but rather refer to a generic wimp of given mass @xmath22 and annihilation cross section @xmath24 dominated by one single annihilation mode .
if the dm annihilation into fermion is not suppressed , quarks give often the dominant branching ratio .
this is the case for a gauge boson wimp , such as the antiperiodic gauge field in @xcite , and for a majorana fermion like the lightest neutralino in supersymmetric extension to the standard model .
for this reason we choose as a benchmark annihilation mode a quark
antiquark pair , giving raise to soft spectra of secondary particles mainly through the hadronization into pions ( charged or neutral ) and their subsequent decay , see fig .
[ fig : yield]a .
the case of a leptonic final state , such as @xmath127 , is rather different since much harder spectrum is produced .
we consider the @xmath94 and @xmath96 as limiting cases of a much more generic wimp scenario . to start our discussion on multi frequency constraints on the gc as a wimp dm source we first focus on three benchmark cases .
properties of the model are listed in table [ tabbm ] and regard the particle physics setup as well as the dark matter profile and its reshaping by the baryonic component in the galactic center region and the assumptions on the magnetic field profile , whose relevance is illustrated in what follows in the discussion of propagation .
accumulate since the two effects balance each other .
the dotted line is the accretion radius @xmath128 , where advection is assumed to stop.,width=294 ] accumulate since the two effects balance each other .
the dotted line is the accretion radius @xmath128 , where advection is assumed to stop.,width=294 ] to model the propagation of electrons / positrons at the galactic center , we need to consider two regimes . outside the accretion flow ,
i.e. at radii greater than the accretion radius @xmath129 pc , the electrons / positrons , injected by dark matter annihilations , lose energy in place through radiative processes and their equilibrium number density is simply given by eq .
[ eqdnde ] ( we will now include all relevant radiative processes ) . for @xmath130 , the physical picture is as follows : the dark matter annihilations inject @xmath131 and @xmath132 at a given radius of injection @xmath133 ; then two competitive processes take place . on top of the momentum loss due to radiative processes , electrons and positrons gain energy in the adiabatic compression due to
the plasma flow onto the central bh .
the propagation equation eq .
[ eqtransp ] admits an integral analytic solution only in case synchrotron emission is the dominant radiative loss process and the @xmath32 are in the ultra
relativistic ( or non relativistic ) regime .
the solution takes the form @xcite : f(r , p)=_r_acc^r dr_inj()^4 c _ ( ) ^4 [ eqodesolrel ] where @xmath134 with @xmath89 being the exponent in the power law scaling of the radial infall velocity @xmath135 , i.e. @xmath65 in case of potential dominated by the central bh , see eq .
[ eqgasvel ] .
the momentum @xmath136 is the initial momentum of an electron injected at @xmath133 , arriving at position @xmath21 with momentum @xmath33 . outside of the ultra
relativistic approximation , eq . [ eqtransp ] can be solved numerically through a change of variables that recasts the original partial differential equation ( pde ) into an ordinary differential equation ( ode ) .
this is defined by a solution of the associated homogeneous equation ; the characteristic curve related to the latter is = , p(r_inj)=p_inj [ eqtraj ] which describes the trajectory of the electrons in the plane radius versus momentum , where p_syn = c_t , p_adv=-(r^2v)p .
[ eqdotp ] the solution of eq .
[ eqtraj ] is shown in fig .
[ fig : traj ] in the plane @xmath137 , in the case of equipartition ( left panel ) or reconnection ( right panel ) magnetic field ( see fig . [ fig : bfield ] ) . in the first case the synchrotron loss dominates at high energies , while the advection gain takes over at low energies ; electrons accumulate on the trajectory separating the two regimes ( black curve in the figure ) .
since approaching the bh , the scaling in radius of the synchrotron loss is faster than the advection gain , @xmath138 versus @xmath139 , the advection dominated region becomes smaller and smaller and disappears for radii very close to the bh horizon . as stated above , in the region with @xmath140 we neglect the advection and thus the trajectories are just horizontal lines . quite similar
is the electron / positron flow in the case of a reconnection magnetic field .
since now the magnetic field is smaller , the advection dominated region becomes larger .
the line along which electrons accumulate is modified accordingly to the shape of the magnetic field plotted in fig .
[ fig : bfield]a .
we would assume a magnetic field which is constant in the accretion region , see again fig .
[ fig : bfield]a , advection basically dominates throughout the plane and there s no region of accumulation .
moreover electrons could be accelerated at energies greater than m@xmath141 , something which is not possible in the previous cases where the propagation of electrons with energy @xmath142 gev becomes dominated by the synchrotron loss at all radii .
we can then solve the propagation equation eq .
[ eqtransp ] on these trajectories , reducing the pde to a linear ode that admits a standard integral solution : f(r , p)=_r_acc^r dr_inj , [ eqodesolut ] where @xmath143 . in the ultra
relativistic limit eq .
[ eqodesolut ] reduced to the form in eq .
[ eqodesolrel ] . at 90 cm for the three benchmark models .
the sharp transition is in correspondence to the accretion radius @xmath128 , where advection starts . in the upper
dashed dotted curves the effect of advection is neglected.,width=294 ] at 90 cm for the three benchmark models .
the sharp transition is in correspondence to the accretion radius @xmath128 , where advection starts . in the upper
dashed dotted curves the effect of advection is neglected.,width=294 ] examples of the resulting electron / positron equilibrium number density are plotted in fig .
[ fig : adv]a .
we can see that the effect of the advection is to drive low energy electrons to higher energies , where synchrotron loss is dominant .
thus there is a peak in the distribution corresponding to the curves of momentum accumulation in fig .
[ fig : traj ] . note that in the case of equipartition magnetic field the accumulation flow is much more efficient with respect to the reconnection case , or , in other words
, there is a wider region of the initial condition @xmath144 for a point of accumulation @xmath137 , and thus more electrons contribute .
for this reason the peak in the density are more pronounced in the equipartition case . in fig .
[ fig : adv]a we plot for comparison the electron / positron equilibrium number density obtained neglecting the effect of advection .
the synchrotron losses dominate until very low energies ( and not too small radii ) where ionization takes over ( see fig .
[ fig : enloss]a ) and the distribution becomes flatter .
[ fig : adv]b gives a feeling for the radial reshaping of synchrotron signals due to advection effects .
we plot the synchrotron luminosity , see eq .
[ eqlumin ] , per unit logarithmic interval @xmath145 , at the wavelength of 90 cm and for the three benchmark models in table [ tabbm ] .
there is a sharp jump in the emissivity at the accretion radius @xmath146 since we have assumed a sharp transition between the two propagation regimes ; in a more realistic model we would find a slightly smoother behavior without , however , prediction for signals significantly affected . at this frequency
the source is rather extended , as already pointed out with the approximate scalings in section 3 .
actually , advection reduces even further the signal from the innermost region .
indeed at large wavelengths the synchrotron power peaks at low energy , while advection shifts electrons from low to high energies .
this effect is more evident for constant and reconnection magnetic fields where the region in the plane @xmath137 dominated by advection is large . for shorter wavelengths ,
the advection effect becomes less and less important since the synchrotron power peak shifts to high energies and thus into the region of the plane @xmath137 in fig .
[ fig : traj ] where the synchrotron losses are dominant .
ray bands for the benchmark models b1 and b2 . in the _ left panel _ ,
an ideal infinite angular resolution is considered , while in the _ right panel _
the intensity is filtered over typical angular resolutions : @xmath147 arcsec at @xmath148 cm , @xmath149 arcsec at @xmath116 hz , and @xmath150 at @xmath93 gev .
dotted lines are the related experimental angular profiles of a point - like source , as modeled by a gaussian detector response.,width=294 ] ray bands for the benchmark models b1 and b2 . in the _ left panel _ ,
an ideal infinite angular resolution is considered , while in the _ right panel _ the intensity is filtered over typical angular resolutions : @xmath147 arcsec at @xmath148 cm , @xmath149 arcsec at @xmath116 hz , and @xmath150 at @xmath93 gev .
dotted lines are the related experimental angular profiles of a point - like source , as modeled by a gaussian detector response.,width=294 ] indirect detection of dark matter through the identification of a photon excess is not a straightforward task .
there is essentially two types of signal for such flux : spectral signatures or signatures related to the morphology of the source .
regarding the spectral signatures , prompt annihilation into monochromatic photons is the most favorable case , however it is not guaranteed in a generic wimp model @xcite . on the other hand ,
signals with continuum energy spectrum could be in general mimicked by standard astrophysical sources .
the spatial structure of the dm source , in case this is extended , could be an equally powerful way of disentangling the source from an environment in which other astrophysical sources are present .
one often has to face the problem that although the wimp source is extended , it can not be experimentally resolved . in the following we want to show that this is not the case at the radio frequencies , since as expected from the approximate results in section 3 the dm source may be very extended .
to study the angular profile of the photon source induced by wimp annihilations , we define as ideal radiation intensity @xmath151 the signal in a detector with an infinite angular resolution . for @xmath18rays ,
the spatial extension is completely fixed by the halo profile , i.e. by the dimension of the dm source .
for synchrotron emission , on the other hand , it is affected by many ingredients , both related to the dark matter properties , to the magnetic field shape , and to the frequency of observation , as we can see from eq .
[ eq : nulnusg ] . in case synchrotron loss is not the dominant radiative loss , also gas and starlight spatial distributions contribute to set the shape of the angular profile of the signal .
we expect from the approximate treatment the radio band signal to become wider than that in the @xmath18ray band , while in the x ray band the need of a very large magnetic field shrinks the signal to a region which is much smaller than the size of the dm source .
this is confirmed in fig .
[ fig : iang]a , where we plot the intensity as a function of the angular off set from the gc , for the benchmark models b1 and b2 in table [ tabbm ] at the radio , x and @xmath18ray bands , normalizing each of the fluxes to unity to better understand the relative spatial extension .
the difference in the spatial extension between the two benchmark models is essentially given by the halo profiles , since the @xmath123 profile leads to a more narrow signal than @xmath122 . in fig .
[ fig : iang ] we are evaluating and including synchrotron self absorption effects , i.e. taking into account that the emitted synchrotron radiation could be reabsorbed by the radiating electrons along the line of sight as described by eq . [ eqdiffisynch ] .
being @xmath152 the synchrotron self absorption coefficient , see e.g. @xcite , the quantity which is useful to estimate the relevance of the absorption effect is the optical depth : _ ( ) = _ losds(,s , ) . in fig .
[ fig : optdepth ] we plot the optical depth along three different lines of sight for the benchmark models b1 and b2 .
as we can see , the absorption effect is relevant only along the lines of sight pointing towards the very central region .
this is due to the fact that the probability of the radiation to be reabsorbed is related to the compactness of the source .
thus in general we expect negligible effects for shallow profiles .
the scaling of absorption with frequency , in general , takes the approximate form : @xmath153 @xcite .
more precisely for the benchmark models , we find numerically that absorption modifies by a factor o(1 ) the flux associated to observations of the inner region in the radio band , while it is irrelevant at larger angles and frequencies . in a real observation ,
the detected angular profile is a combination of the intrinsic profile shown in fig .
[ fig : iang]a and the experimental resolution , as described by eq . [ eqisynch ] .
in fig . [
fig : iang]b we plot again the wimp induced emissions for the benchmarks models b1 and b2 , now filtered over a typical angular resolution .
for the 90 cm signal , we take a typical resolution achievable by vla , namely fwhm=@xmath154 @xcite . for the x rays emission we consider the chandra point spread function , i.e. psf=@xmath155 @xcite .
finally in the @xmath18ray case , the signal is integrated over @xmath150 that is a typical psf for both the upcoming gamma - ray telescope in space glast @xcite and the current generation of act @xcite . the synchrotron emission in the x
band is very narrow and thus impossible to be resolved . in case of a very cuspy profile , like @xmath123
, the source could be resolved only by radio observations , while for the @xmath122 profile the source could be detected as diffuse both in the @xmath18ray and radio bands . in fig .
[ fig : inub]a we plot the radiation intensity for the benchmark model b1 at four different frequencies .
as expected , the size of the source becomes smaller going from radio to infrared wavelengths . in fig .
[ fig : inub]a , we show also the angular resolution of the current or near future experiments in the radio ( vla @xcite and evla @xcite ) , mm and sub - mm ( alma @xcite ) , infrared and nir ( vlt @xcite ) bands , relative to the wavelength scale plotted on the right - hand side . in the first two cases the wimp source appears extended ,
while going to higher frequencies it becomes a point source . in fig .
[ fig : inub]b , we show the effect of varying the magnetic field on the synchrotron intensity at 90 cm , for the benchmark models b1 and b2 , but looping over the magnetic field shapes shown in fig .
[ fig : bfield]a .
note that the choice of magnetic field differs only inside the accretion region @xmath156 .
equipartition choice gives the brightest signal , while the constant magnetic field the faintest . at this wavelength and for typical angular resolution of current detectors ,
the contribution from the region @xmath157 is never dominant , hence the choice of the magnetic field is essentially not relevant . the fact that at the intermediate angular scale the prediction of the two models coincides reflects just the choice of normalizing their radio emissivity to the tightest upper bound in the radio band , i.e. @xmath6 jy at the frequency @xmath7 mhz @xcite . for a given magnetic profile , the higher the energy for the radiating electron or positron , the higher the frequency at which the synchrotron power peaks ; thus a hard @xmath32 spectrum emits more efficiently at short wavelengths , while a soft spectrum at long wavelengths . in fig .
[ fig : ifsrho]a we plot the angular intensity of the synchrotron signal in the radio , nir , and x bands for benchmark model b1 and b4 , i.e. the same of b1 except for assuming @xmath96 as dominant annihilation channel rather than @xmath158 .
we find that b4 is significantly brighter than b1 at high frequencies .
note also that the spatial extension at such frequencies depends significantly on the wimp annihilation final state .
finally we turn to the uncertainty on the dimension of the signal stemming from the dimension of the source itself . in the case of signals at radio frequencies
the scale at which is relevant to model the dm density profile to derive a definite prediction correspond basically to the angular resolution of the observation itself , i.e. @xmath159 . in fig .
[ fig : ifsrho]b , we plot the benchmark model b1 at 90 cm , varying the dark matter profile and find how dramatically the signal can change .
note that the reason why the result with the nfw profile or the @xmath122 profile are essentially equivalent is the large value of the ratio @xmath160 for the benchmark model under consideration , which is flattening out the @xmath122 profile .
cm , but varying the magnetic field according to the three radial profiles in fig.[fig : bfield]a ( same line styles).,width=309 ] cm , but varying the magnetic field according to the three radial profiles in fig.[fig : bfield]a ( same line styles).,width=294 ]
in the previous section , we discussed how intensity and spatial extension of the signal depend on parameters involved in the prediction for the multi wavelength spectra .
we implement now this general analysis to derive quantitative constraints . in section 2
we listed available data
sets on the gc relevant in our analysis ; since it is unlikely that any of them comes in connection to a dm signal , we will extract upper limits only .
it s not straightforward to select a uniform exclusion criterion for all the measurements .
we decide to compare the dm induced flux with the most constraining data point in any given wave band . to some extent
, this is a conservative approach , since each experimental point is considered to be independent and no correlation analysis implemented . in the next decade , new telescopes , as well as new cycles of observations with experiments already operative at present , will allow to place even tighter constraints on wimp parameter space , or , hopefully , find evidence for a dm signal .
we will focus , in particular , on two classes of @xmath18ray telescopes , namely , the satellite detector glast @xcite which will be in orbit in a few months , and the next - generation air cherenkov telescope cta @xcite , and discuss the relevance of new observations at radio frequencies by the vla project @xcite . as summarized in the section 2 ,
rather accurate measurements of the radio and infrared emission of the source associated to the central smbh are available .
both the spectrum and the pattern in size of this source can not be associated to synchrotron emission from dm annihilations .
typically , observations of sgr a@xmath0 have been obtained with instruments with very good pointing accuracy and small angular acceptance . on the other hand ,
wimp annihilations give rise to radio signals on a much larger angular size .
it follows that , in general , it is incorrect to directly compare the total radio luminosity of the dm source with the luminosity extrapolated from the available sgr a@xmath0 observations .
a more accurate way of proceeding is to compute , for each model and each data point , the dm induced synchrotron intensity within the region corresponding to the angular resolution of the telescope , i.e. mimicking a gaussian response of the detector with @xmath76 in eq .
[ eqisynch ] ( or a gaussian elliptical response with two different @xmath28 ) as appropriate for each measurement . in fig .
[ fig : snu90]a we show measured intensities ( or upper limits ) for sgr a@xmath0 ( @xcite , @xcite , @xcite ) together with the dm synchrotron luminosity @xmath161 integrated over the whole gc region , say , e.g. , a sphere of radius corresponding to an angular size of about @xmath162 , and divided by @xmath163 , where @xmath164 is the distance to the gc ( solid , dashed , and dotted lines , respectively , for benchmark models b1 , b2 , and b3 , spanning the whole range of frequencies shown in the plot ) . as we just stated , this is not the quantity which should be compared to radio data ; would one make such a connection , i.e. implicitly assuming that the dm source is point - like rather than extended , the inferred upper bounds would be grossly overestimated .
we select instead five data - points ( plus one in the infrared ) , each corresponding to measurements with different angular resolutions , and plot , in a small interval around the corresponding frequency , intensities towards the gc , treating now the signal as an extended source filtered by the telescope angular response .
as expected , the strongest constraint in the radio band comes from the measurement at the lowest available frequency @xcite and the value of the cross sections for the benchmark models have been tuned to match this upper limit .
this is also the measurement we will refer to , when combining constraints from different frequencies to the multi
wavelength dm spectrum in figs .
[ figsvvsmb ] and [ figsvvsmtau ] below . as dominant annihilation channel ) .
we show the signal at different wavelengths , namely , in the radio , nir , and x ray bands . _
right panel _ : angular profile of the synchrotron flux intensity at 90 cm for the benchmark model b1 , but varying the dm halo profile.,width=294 ] as dominant annihilation channel ) .
we show the signal at different wavelengths , namely , in the radio , nir , and x ray bands . _
right panel _ : angular profile of the synchrotron flux intensity at 90 cm for the benchmark model b1 , but varying the dm halo profile.,width=294 ] the intrinsic dimension of the dm synchrotron source at radio frequencies suggests that observations covering a wider field of view could set relevant constraints as well .
we consider the map of the galactic center at @xmath165=90 cm obtained by @xcite , assembling different vla observations .
it is a @xmath166 image , with a resolution of @xmath167 , thus resolving sgr a , the complex radio source present at the gc and composed by sgr a@xmath0 , the supernova remnant sgr a east , and the spiral structure sgr a west , but not sgr a@xmath0 itself .
the background noise level is about 5 mjy / beam . in fig .
[ fig : snu90]b we plot the radial profile of the dm signal as it would be reported in a map with the resolution of @xcite and detected by an observation with a resolution of @xmath168 , like in the sgr a@xmath0 survey of @xcite .
the sgr a source is not spherically symmetric and its angular profile can not be accurately reduced to a radial profile ; in fig .
[ fig : snu90]b we give just a schematic representation of the angular shape of the signal reported by @xcite .
we find that the limit on dm models one can deduce from sgr a data is less stringent than the constraint inferred from sgr a@xmath0 . at large angles , however , the dm signal is comparable to the background noise level , in particular , in the case of the @xmath122 profile . as mentioned above ,
such noise level is extrapolated in @xcite , assembling observations with different resolutions .
it is not the best achievable in vla observations today , of the order of @xmath169 at 90 cm @xcite .
new wide - field observations could indeed lead to tighter constraints on dm models , as we will be discuss below . in fig .
[ fig : snu90]a we plot three measurements of the nir luminosity of sgr a@xmath0 in the quiescent state @xcite , plus three upper limits derived in @xcite , and the dm induced signal for the three benchmark models .
we discussed in some detail how the angular size of the source shrinks rapidly going to higher frequencies .
for the halo profiles we consider in our analysis , already in the nir the dm source would appear as point like , even with a detector with excellent angular resolution , such as a size of tens of mas achievable by vlt @xcite .
indeed , one can see that the estimate of the signal computing @xmath170 , or @xmath171 integrated over the appropriate angular size , essentially coincide .
measurements are not far above from the estimated dm luminosities , especially for the benchmark model b2 , for which this limit is comparable to the radio limit .
we will derive limits on wimp masses and cross sections considering the tightest nir limit , namely , the measured emission in the @xmath172 band ( @xmath173 ) .
luminosity in the radio and infrared bands ( black points ) compared to the synchrotron radiation induced by dm annihilations in the benchmark models b1 , b2 , and b3 . for the latter , portions of spectra integrated over the experimental angular resolutions around six frequencies
are shown .
the lines spanning the entire range of frequencies are the spectra integrated over the whole gc region . _
right panel : _ spatial profile of the dm synchrotron signal for the benchmark models b1 , b2 , and b3 . in the upper curves
the angular resolution is @xmath167 , while in the lower curves it is @xmath168 .
we show together the experimental limits related to the galactic center region derived by @xcite and to sgr a@xmath0 by @xcite.,width=294 ] luminosity in the radio and infrared bands ( black points ) compared to the synchrotron radiation induced by dm annihilations in the benchmark models b1 , b2 , and b3 . for the latter , portions of spectra integrated over the experimental angular resolutions around six frequencies are shown .
the lines spanning the entire range of frequencies are the spectra integrated over the whole gc region . _
right panel : _ spatial profile of the dm synchrotron signal for the benchmark models b1 , b2 , and b3 . in the upper curves
the angular resolution is @xmath167 , while in the lower curves it is @xmath168 .
we show together the experimental limits related to the galactic center region derived by @xcite and to sgr a@xmath0 by @xcite.,width=294 ] significant synchrotron emission at even higher frequencies is expected in the case of very large magnetic fields close to the central black hole , as in the equipartition and reconnection magnetic field models we are considering . for the flux emitted in the uv and soft x band , we need to take into account the attenuation due to the photoelectric effect on the interstellar dust .
we model this effect scaling down the emissivity of eq .
[ eqjsynch ] by the factor @xmath174 , where @xmath175 is the electron column density @xcite and @xmath176 is the photoelectric cross section @xcite . in fig .
[ fig : snux]a , we plot the dm signal due to synchrotron emission , in the energy range where chandra @xcite has detected an x ray source with position consistent with sgr a@xmath0 .
the three benchmark models are considered , as well as the cases in which , keeping all the other parameters in the model fixed , the other choices for the magnetic field radial profile ( see fig . [
fig : bfield]a ) are implemented . to convert flux intensities into counts per unit energy and time
, we use the chandra effective area on axis reported in @xcite . for a wimp with mass of about @xmath93 tev ( upper green curves )
the peak in the emissivity is at galactocentric distances at which equipartition and reconnection magnetic fields differ only slightly , and thus the relative signals do not differ dramatically . in the case of the magnetic field flattened to a constant value ( dotted green curve ) , on the other hand , synchrotron emissivity is sharply suppressed .
for @xmath177 gev wimps ( blue and red curves ) , the signal originates in a much smaller region , where equipartition and reconnection magnetic fields differ substantially , and the constant magnetic field can not give a sizable signal . to better understand the dependence on the wimp mass of the synchrotron signal , we show the x
rays spectrum in fig .
[ fig : snux]b for the benchmark models , and consider three wimp mass scales .
a. the black line is the fit to the chandra measured spectrum .
_ right panel : _ synchrotron x ray spectrum originated from dark matter annihilations in the benchmark models b1 , b2 , and b3 , but varying the mass .
the three cases considered are ( from top to bottom ) : @xmath93 tev , @xmath177 gev , and @xmath59 gev . for constant magnetic field only the first case is shown since smaller masses can not give a sizable spectrum.,width=275 ] a. the black line is the fit to the chandra measured spectrum . _ right panel : _ synchrotron x ray spectrum originated from dark matter annihilations in the benchmark models b1 , b2 , and b3 , but varying the mass .
the three cases considered are ( from top to bottom ) : @xmath93 tev , @xmath177 gev , and @xmath59 gev . for constant magnetic field only the first case is shown since smaller masses can not give a sizable spectrum.,width=294 ] at x - ray frequencies and above , the dominant radiative process involving the @xmath32 produced by wimp annihilations can be inverse compton scattering , rather than synchrotron emission .
ic on the cosmic microwave background is peaked in the x band , while that on the starlight has its peak in the multi mev or even gev region .
the distribution of starlight in the galaxy is highly non uniform ; its average energy density in the inner region is about @xmath178 ev @xmath42 @xcite . as a sample ansatz to make an estimate of the level of ic emission on starlight , we assume that such a value can be representative for the whole gc region and for simplicity we will make also the approximation of the starlight spectrum black body shape of temperature @xmath179 ev @xcite . in fig .
[ fig : sicgam ] , we plot the ic spectra on cmb and starlight , induced by wimp
annihilations in the three benchmarks models .
it is shown for a typical angular resolution of the current @xmath18rays experiments , i.e. @xmath180 sr .
we are considering such a large field of view since the ic signals have an angular shape which is significantly broader than the shape of the @xmath32 source function .
we can intuitively understand this feature from the fact that this emission comes mostly in connection to the @xmath32 with largest energy at emission , and these in turn lose energy by synchrotron losses much more efficiently close to the gc , where magnetic fields are the largest , than in the outskirts of the gc region .
it turns out that the angular shape for the equilibrium number density of high energy @xmath32 is much broader than the gamma - ray flux from @xmath181 decays ( which is the same as for the source function ) , and , of course , even more with respect to the shape of the synchrotron induced x - ray flux . for this reason , although for the plot in fig .
[ fig : sicgam ] the intensity associated to the ic on cmb is larger than the synchrotron intensity , when integrating over the angular resolution of the chandra detector , the trend is reversed , and only in the case of constant magnetic field , with synchrotron emission in the x - ray band essentially negligible , comparing the ic flux to sgr a@xmath0 gives a tighter constraint .
analogously to what we did in the case of radio emission , it is worth checking whether data on a large field of view could be relevant .
we compare the ic signal to the diffuse x
rays emission detected by the chandra observatory : in the @xmath182 map of @xcite , some regions are selected and from them spectra of diffuse emission are extracted , removing events near points source and filamentary features . when combining constraints from different frequencies in figs . [ figsvvsmb ] and [ figsvvsmtau ] below , we compute the level of ic emission in such regions and extract upper bounds .
similar arguments apply for the ic on starlight and the @xmath18ray limits . indeed for what concerns bounds associated to the point like source detected by egret at the gc ( actually its position is controversial , see the next section ) , the limit associated to @xmath181 decay is more constraining than the ic limit .
this is not true in general for the diffuse emission on the whole gc region , however we do not find any region in the parameter space in which tighter limits come in connection to this component .
note that the assumption we made on radial profile and energy spectrum for the starlight background are rather crude , and may deserve further study ; refining them may lead to a slightly different conclusion , but it is unlikely that the general picture would be affected .
recently , observations by atmospheric cherenkov telescopes detected a gamma - ray source in the direction of the galactic center .
in particular the h.e.s.s .
collaboration ( @xcite , @xcite ) has obtained an accurate measurement of the spectrum of the source as a single power law in the energy range between 160 gev and a few tens of tev , making the interpretation of the signal in terms of wimp dm pair annihilations rather unlikely .
h.e.s.s . has found evidence for a gc point
like source , namely , a source with an extension smaller than its psf=@xmath150 and position compatible with sgr a@xmath0 , on top a diffuse @xmath18-ray component @xcite . in the case of cuspy dark matter halo profiles , one needs to compare against the central source only ; the shallower the profile , the more efficient it becomes to extend the analysis and include the gc ridge as well ( see , e.g. , the discussion in @xcite ) .
the resulting limits for the benchmark profiles are plotted in figs .
[ figsvvsmb ] and [ figsvvsmtau ] .
ray emissions induced by dm annihilations for the benchmark models b1 , b2 , and b3 .
all the four mechanisms of photon spectrum production considered in the paper give sizable signals .
the flux intensities are integrated over a solid angle of @xmath180 sr .
the level of the diffuse emission detected by chandra is also shown ( black line).,width=10 ] the egret telescope mapped the gc in the energy range 30 mev10 gev @xcite , detecting a flux within @xmath183 of the gc . a few hypothesis for interpreting this flux in terms of a standard astrophysical source
have been formulated ; its spectral shape is even compatible with a component from wimp dm annihilations @xcite . on the other hand ,
the poor angular resolution of egret does not allow for a univocal identification of the source . in ref .
@xcite , using only energy bins above @xmath93 gev and a spatially unbinned maximum likelihood analysis , the authors argue that the galactic center is excluded as the position of the source at @xmath184 and the maximum likelihood location is at @xmath185 , @xmath186 .
thus they derive upper limits on the @xmath18rays flux from dm annihilations under the condition of no evidence of a point source at the gc .
whether this is the correct approach is still under debate and only glast will give a definitive answer .
we derive more conservative but robust limits comparing with the egret source ; would one follow the line of @xcite , the limits would be improved up to about a factor of ten . except for very light wimps ,
the strongest constraint comes from the last data - point in the egret measurement , in the energy bin @xmath187 gev .
having specified how individual constraints are implemented , we are now ready to discuss the global picture .
we refer to a model independent scenario in which a wimp model is labeled by the value of the wimp mass @xmath188 and its total annihilation rate @xmath24 , both assumed as free and independent parameters . as for the benchmark cases , we discuss as extreme cases for the wimp source functions , a soft spectrum configuration fixing to 1 the branching ratio in the @xmath94 channel , and a hard spectrum setup when @xmath96 is the dominant annihilation mode . again , having specified the annihilation mode and the wimp mass , injection spectra are fixed accordingly to simulation results with the pythia package as implemented in darksusy @xcite .
reference models for the dm distribution in the gc region are the @xmath122 and @xmath123 profiles ( with the second much denser than the first , hence with upper bounds on @xmath24 expected to shift dramatically ) . finally , we loop over the three reference magnetic field radial profiles given in fig . [
fig : bfield]a . in figs .
[ figsvvsmb ] and [ figsvvsmtau ] we consider the four possible combinations of annihilation channels and halo profile .
the davies et al .
radio bound does not depend on the magnetic field choice since , as we have seen above , the signal is generated mainly outside the accretion region .
the same is of course true for the egret and hess @xmath18-ray limits .
it is rather striking to see that the radio limit is always tighter than the egret limit , with this trend getting enforced even more , the softer the spectra and the more cuspy the halo profile .
were we considering a dm profile obtained by implementing the original simplified procedure by gondolo and silk as response for the adiabatic formation of the central smbh @xcite , we would find that essentially the whole wimp parameter space is excluded , as in the original conclusions in refs .
@xcite ( despite the fact that several ingredients in this analysis are refined and/or treated differently ) .
the hess limit becomes more stringent for heavy wimps , especially in the case of hard emission spectra .
unfortunately this is a regime in which other constraints take over . as dominant annihilation channel .
the _ left panel _ and _ right panel _ show the limits for , respectively , the @xmath122 and @xmath123 profiles ; note the mismatch on the vertical scale in the two plots . the radio constraints from davies et al . , the limits from egret and hess @xmath18-ray measurements , and the bound from the x - ray diffuse emission as detected by chandra ( dashed dotted line ) , do not depend on the choice of the magnetic field radial density profile .
the constraints associated to the nir and x - ray observations of sgr a@xmath0 , respectively , by vlt and chandra , are shown for the three magnetic field models of fig .
[ fig : bfield]a ( using the same line styles ) . , width=294 ] as dominant annihilation channel .
the _ left panel _ and _ right panel _ show the limits for , respectively , the @xmath122 and @xmath123 profiles ; note the mismatch on the vertical scale in the two plots . the radio constraints from davies et al .
, the limits from egret and hess @xmath18-ray measurements , and the bound from the x - ray diffuse emission as detected by chandra ( dashed dotted line ) , do not depend on the choice of the magnetic field radial density profile .
the constraints associated to the nir and x - ray observations of sgr a@xmath0 , respectively , by vlt and chandra , are shown for the three magnetic field models of fig .
[ fig : bfield]a ( using the same line styles ) .
, width=294 ] vlt nir limits depend to some extent to the magnetic field choice and show some non trivial behavior .
consider the case of the @xmath96 final state .
for very heavy wimps , and hence very energetic radiating particles in this hard emission spectra , the value of the magnetic field matching the peak in synchrotron emission is quite small ( see fig . [
fig : bfield]b ) , corresponding to the region where we have assumed identical shapes for the profile of the three benchmark cases .
going to smaller masses , the energy at which the @xmath32 distribution peaks becomes smaller , and thus the required magnetic field higher , approaching the value we assigned ( by mere chance ) to the central plateau in the constant magnetic field case ( limits are coded in magnetic field using the same convention for line - type as in fig .
[ fig : bfield]a ) ; the emission is particularly efficient and bounds are more effective with respect to the equipartition and reconnection magnetic field cases . at smaller masses the magnetic field matching the synchrotron peak becomes greater than the constant plateau and constraints are quickly relaxed .
the same effect happens for the reconnection magnetic field , at even smaller masses .
an analogous effect takes place for the @xmath94 channel , but to a smaller extent due to the soft spectrum .
we have already discussed patterns of dependences of the synchrotron x - ray signal with the magnetic field in many details . for moderate to large values of magnetic fields around the central bh ,
the limit from the detection of sgr a@xmath0 by chandra tends to be the tightest in the wimp parameter space , except if the wimp mass is too small , the annihilation channel is too soft , or the density of wimp very close to the gc is not large enough , i.e. if , in connection to one or more of these issues , we do not have enough high energy radiating electrons and positrons . the signal is generated in a very small region , where the dm profile depends on the ratio @xmath189 , and hence the scaling of the flux with the cross section is not linear . in the case of the @xmath123 profile , this dependence is so strong that the limit can be double valued .
finally , the dash - dotted line refers to the limit extracted from detection by chandra of a diffuse x - ray background , when compared to the predicted ic emission on the cmb . it can be the tightest x - ray limit , however , it is never the strongest constraints in any combination of our reference setups .
in general , the request for the wimp thermal relic abundance to not exceed the value of the mean dm density in the universe as derived from cosmological measurements , fixes a lower bound on the total annihilation rate at zero temperature ( the relic density scales approximately with the inverse of the pair annihilation rate ; there are , however , cases when such correspondence is badly violated , the prime example being when coannihilation effects are present ) .
the very tight constraints we have found in case of the @xmath123 profile should make very narrow , or even close , the allowed window in the wimp parameter space . for the @xmath122 profile , on the other hand , the limits we have derived are much less stringent . , but
taking @xmath96 as the dominant annihilation channel.,width=294 ] , but taking @xmath96 as the dominant annihilation channel.,width=294 ] indirect dark matter detection is one of the most ambitious objectives for new observational campaigns or new telescopes getting available in the near future , with the gc often being indicated as the prime observational target .
we try to make here a projection on how significant could be the improvement with respect to the region of the wimp parameter space already excluded in figs .
[ figsvvsmb ] and [ figsvvsmtau ] .
we mentioned that the radio bounds could become even stronger for wide field @xmath148 cm observations of the gc region reaching a noise level which is significantly reduced with respect to the map constructed in @xcite , at least in case the intrinsic dimension of sgr a in the radio band is not much larger than what is inferred from present observations . in figs .
[ figsvvsmprojb ] and [ figsvvsmprojtau ] we sketch the case of a hypothetical observation with the vla in its configuration with the worst angular resolution for spatial reconstruction , but with the maximal dm - signal to noise ratio , namely with fwhm=@xmath190 and a noise level of @xmath191 mjy in @xmath192 hours of observations ( configuration d in ref .
we are pointing the telescope at an angle of @xmath193 with respect to the gc .
the lower curves sketch the improvements in upper bounds which could be obtained in the case of presence of regions with no contaminations from astrophysical backgrounds ( 3 @xmath92 noise level ) .
this scenario corresponds to the most favorable case .
indeed a 90 cm diffuse emission at the gc was already detected @xcite .
however , the poor angular resolution of the surveys ( 51 and 39 arcmin , respectively ) does not allow to derive the spatial structure of the emission in the innermost region . in the gc image of @xcite , such emission does not seem completely isotropic and hence , from patches of the map with no background , we can extract tighter bounds ( upper curves in figs .
[ figsvvsmprojb ] and [ figsvvsmprojtau ] ) with respect to @xcite . indeed , although the observations made with the green bank telescope and reported in @xcite are consistent with the bound derived in @xcite by the comparison of the detected diffuse non - thermal source and the expected synchrotron emission from gc cosmic rays . ]
have a comparable sensitivity , the associated image shows a smoother diffuse emission , due to the larger angular resolution , and the comparison between the wimp signal and the noise level has to be performed at larger angles , where the dm emission is fainter .
the real limit is probably standing in between the two extreme cases plotted in figs .
[ figsvvsmprojb ] and [ figsvvsmprojtau ] .
the space satellite glast has been launched very recently and will be soon operative .
the energy range of detection is approximately 100 mev300 gev , with an expected sensitivity improved by a factor @xmath177 with respect to egret .
the psf and the effective area at high energy are respectively @xmath180 sr and @xmath194 @xmath195 ( in the following we will consider the full energy dependence in these quantities as inferred from @xcite ; averaging over the angle of observation at which the gc stands with respect to the zenith of the detector are included as well , finding an effective exposure which is essentially reduced by a factor of 0.3 ) .
we have also assumed a @xmath196 energy resolution , an exposure time of @xmath197 years , and systematic errors of 5.2@xmath198 @xcite ; the latter are relevant only at energies @xmath199 gev . the next generation of act , the cherenkov telescope array ( cta ) project , is currently under development .
the proposed energy range of detection is 10 gev100 tev , thus overlapping and extending on the hess range .
the most dramatic improvement will be in the effective area , up to about @xmath93 km@xmath200 or even larger in extreme configurations , with highly reduced statistical errors .
based on the study in @xcite , we assume systematic errors to be @xmath2011% , the energy resolution at the level of @xmath196 , and the point spread function equal to @xmath180 sr . for an act , on top of astrophysical backgrounds , one needs to take into account the background from misidentified showers ,
i.e. : = + [ eqmisshow ] where @xmath202 are the spectra of the gamma - like showers from hadrons and electrons , respectively .
we treat these components following @xcite , assuming @xmath203 of misidentified hadron showers with respect to the total incident flux .
we assume a total of about @xmath204 hours for the exposure time ( reasonable in 5 years of operation for cta ) .
to estimate the @xmath18ray projected constraints in the plane dm mass versus annihilation cross section , we make an extrapolation of the point like and diffuse astrophysical backgrounds detected by hess over the whole energy range of interest , namely 1 gev300 gev for glast and 10 gev100 tev for cta , assuming single power law scaling for the fluxes .
we consider two generic power law spectra @xmath205 , with @xmath206 , one for the point
like gc source and the other for the diffuse gamma
ray emission in the galactic center region , assumed to have a flat angular profile .
we first generate a sample of @xmath207 and @xmath208 coefficients satisfying the condition @xmath209 when compared to the data - sets from hess observations @xcite and @xcite .
then , we simulate how this flux should be seen by glast and cta summing statistical and systematic errors in quadrature ( we define the statistical error as the square root of the number of events in any given bin ) . finally , we compute the best fits assuming as theoretical flux a dark matter contribution on top of a new two - component background @xmath210 . among all the @xmath211 and @xmath212 coefficients allowed
, we retain the case providing the smallest @xmath213 and take as exclusion criterion @xmath214 , namely a flux not well fitted by the dark component plus any viable astrophysical components .
the @xmath213 analysis is performed both on the energy spectra and on the angular structure of the flux .
the angular bin size is fixed according to the psf . for the @xmath123 profile ,
this last step is useless , since the dark matter signal is concentrated in the central angular bin ( see fig .
[ fig : iang ] ) , while for the less cuspy @xmath122 profile this procedure provides additional information .
( the method we are implementing leads to analogous conclusions with respect to the treatment in @xcite , the main differences in the extrapolated limits stemming from the different halo profiles adopted and a different treatment of systematic errors . ) results are shown in figs .
[ figsvvsmprojb ] and [ figsvvsmprojtau ] . in the same plots ,
shaded regions identify the models violating at least one of the constraints in figs .
[ figsvvsmb ] and [ figsvvsmtau ] considering the weakest limit among the three cases with a different choice of the magnetic field radial profile , i.e. models that are _ excluded _ ( at least within the rather general set of assumptions we are making regarding magnetic fields , treatment of electrons and positrons propagation , dark matter densities in the gc region , and spectral features of the yield from wimp annihilations ) . the projected limit for glast is always lying in a shaded region ; those for cta span modest portions of the parameter space which are not already excluded .
one should consider , on one hand , that we may have been over conservative , since we derived these limits relying on extrapolations on both the energy spectra and the angular profile for the background astrophysical components , as well as without assuming any theoretical modeling of such astrophysical sources ; with data at hand the picture may look slightly more favorable . on the other hand , this is indeed suggesting that , although the @xmath18-ray band is the regime in which it is most straightforward to make the connection between a given dark matter model and the induced signal , it does not seem to be the energy range with the best signal to background ratios , at least in the case of the gc and of not very cuspy dm profile . as the dominant annihilation channel .
the _ left panel _ and _ right panel _ show the limits for , respectively , the @xmath122 and @xmath123 profile .
the glast and cta projections are obtained combining an angular and spectral analysis as described in the text .
the vla limit arises from the comparison with the background noise level at 50 arcmin away from the gc .
the upper curve is derived assuming a noise level as in @xcite , while the lower curve is computed considering the minimal noise achievable by vla ( d configuration ) .
shaded regions identify the models violating at least one of the constraints in fig .
[ figsvvsmb ] ( considering the weakest limit among the three cases with a different choice of the magnetic field radial profile ) . , width=294 ] as the dominant annihilation channel .
the _ left panel _ and _ right panel _ show the limits for , respectively , the @xmath122 and @xmath123 profile .
the glast and cta projections are obtained combining an angular and spectral analysis as described in the text .
the vla limit arises from the comparison with the background noise level at 50 arcmin away from the gc .
the upper curve is derived assuming a noise level as in @xcite , while the lower curve is computed considering the minimal noise achievable by vla ( d configuration ) .
shaded regions identify the models violating at least one of the constraints in fig .
[ figsvvsmb ] ( considering the weakest limit among the three cases with a different choice of the magnetic field radial profile ) . , width=294 ]
we have presented a systematic , self - consistent study of the multi wavelength emission due to wimp pair annihilations in the galactic center region .
the wimp signal is expected to extend from the radio band up to gamma - ray frequencies .
the gamma - ray luminosity is mostly associated to the chain of decays and/or hadronization processes initiated by two - body annihilation channels , leading to the production of neutral pions and their subsequent decays into two photons . in analogous chains , and with comparable efficiencies , high - energy electrons and positrons
are produced as well : emitted in a region with large magnetic fields , they give rise to synchrotron emission covering radio frequencies up to , possibly , the x - ray band .
a minor role is also played by inverse compton scattering on the cosmic microwave background or starlight . referring to a generic wimp dm scenario , we have discussed spectral and angular features , and sketched the correlations among signals in the different energy bands .
we have illustrated which are the critical assumptions in deriving such conclusions , starting from uncertainties in the dm source functions , regarding both wimp models and dm distributions , up to the modeling of propagation for electrons and positrons , and the assumptions on magnetic field profiles .
we have introduced benchmark cases to guide the discussion and extracted the most relevant general trends : radio to mm synchrotron emission is essentially independent from the shape of the magnetic field in the innermost region of the galaxy , while at shorter wavelengths , i.e. in the infrared and , especially , the x ray band , a different choice for the magnetic field may change predictions dramatically .
radio signals have in general very large angular sizes , larger than the typical size for the source function and hence of the @xmath18-ray signals .
the size of the region of synchrotron x - ray emissivity shrinks dramatically going to larger frequencies , smaller wimp masses , or softer annihilation channels .
the luminosity of the wimp source at the different frequencies , and especially comparing the radio to the @xmath18-ray band , is essentially at a comparable level , with luminosity ratios depending rather weakly on wimp mass and annihilation channels .
this is interesting , since the gc astrophysical source sgr a@xmath0 , an unusual source , certainly very different from typical galactic or extragalactic compact sources associated to black holes , has a very low luminosity over the whole spectrum , at a level at which it is plausible that a wimp - induced component may be relevant . indeed , after a closer look , one sees that none of the fluxes detected in gc direction has spectral or angular features typical of a dm source , still all data - sets contribute to place significant constraints on the wimp parameter space .
we have found that , although the @xmath18-ray band is the regime in which it is most straightforward to make the connection between a given dark matter model and the induced signal ( hence it is also the regime on which most of previous analyses have concentrated on ) , it does not seem to be the energy range with the best signal to background ratios . in the case of large magnetic fields close to the gc , x - ray data can give much tighter constraints .
radio and nir measurements , which are less model dependent , tend to be more constraining as well . regarding an outlook for the future
, we have explored the capability of improving @xmath18-ray constraints on wimp models of the glast satellite telescope , and of cta as representative of the next generation of air cherenkov telescopes .
the recent discovery of a @xmath18-ray gc source and of a diffuse @xmath18ray component , however , limits the possibility of dramatic improvements , possibly reducing the region in the parameter space accessible to @xmath18-ray telescopes to regimes that , within the range of assumptions listed in our analysis , are already excluded at other wavelengths . on the other hand ,
if the sgr a source has a size in the radio band which is not significantly larger than its presently estimated value , future wide field radio observations could be a new effective way to test wimp dm models . , but taking @xmath96 as the dominant annihilation channel.,width=294 ] , but taking @xmath96 as the dominant annihilation channel.,width=294 ]
we would like to thank g. bertone and d. merritt for kindly providing some of dark matter halo profiles which have been used in the analysis .
we also would like to thank s. colafrancesco , m. massardi and a. michelangeli for useful discussions .
we thank the anonymous referee for providing comments which helped in improving the present version of the work .
this work is partially supported by the european community s human potential programme under contract no .
mrtn - ct-2006 - 035863 .
g. jungman , m. kamionkowski and k. griest , phys .
* 267 * ( 1996 ) 195 [ arxiv : hep - ph/9506380 ] .
l. bergstrom , rept .
* 63 * ( 2000 ) 793 [ arxiv : hep - ph/0002126 ] .
g. bertone , d. hooper and j. silk , phys .
* 405 * ( 2005 ) 279 [ arxiv : hep - ph/0404175 ] .
g. d. mack , t. d. jacques , j. f. beacom , n. f. bell and h. yuksel , arxiv:0803.0157 [ astro - ph ] .
m. wood _ et al .
_ , arxiv:0801.1708 [ astro - ph ] . m. kachelriess and p. d. serpico , phys . rev .
d * 76 * ( 2007 ) 063516 [ arxiv:0707.0209 [ hep - ph ] ] .
e. j. ahn , g. bertone and d. merritt , phys .
d * 76 * ( 2007 ) 023517 [ arxiv : astro - ph/0703236 ] .
j. f. navarro , c. s. frenk and s. d. m. white , astrophys . j. * 490 * ( 1997 ) 493 [ arxiv : astro - ph/9611107 ] .
j. f. navarro _ et al .
_ , mon . not .
* 349 * ( 2004 ) 1039 [ arxiv : astro - ph/0311231 ] .
j. diemand , m. zemp , b. moore , j. stadel and m. carollo , mon . not .
* 364 * ( 2005 ) 665 [ arxiv : astro - ph/0504215 ] .
r. a. flores and j. r. primack , astrophys . j. * 427 * ( 1994 ) l1 [ arxiv : astro - ph/9402004 ] .
b. moore , nature * 370 * ( 1994 ) 629 .
g. gentile , p. salucci , u. klein , d. vergani and p. kalberla , mon . not .
* 351 * ( 2004 ) 903 [ arxiv : astro - ph/0403154 ] .
r. genzel , c. pichon , a. eckart , o. e. gerhard and t. ott , mon . not .
soc . * 317 * ( 2000 ) 348 [ arxiv : astro - ph/0001428 ] .
a. ghez , m. morris , e. e. becklin , t. kremenek and a. tanner , nature * 407 * ( 2000 ) 349 [ arxiv : astro - ph/0009339 ] .
r. schodel _ et al .
_ , nature * 419 * , 694 ( 2002 ) . f. aharonian _ et al .
_ [ the hess collaboration ] , astron .
astrophys .
* 425 * ( 2004 ) l13 , [ arxiv : astro - ph/0408145 ] .
l. bergstrom , p. ullio and j. h. buckley , astropart .
* 9 * ( 1998 ) 137 [ arxiv : astro - ph/9712318 ] .
a. cesarini , f. fucito , a. lionetto , a. morselli and p. ullio , astropart .
* 21 * ( 2004 ) 267 [ arxiv : astro - ph/0305075 ] .
d. hooper and b. l. dingus , phys .
d * 70 * ( 2004 ) 113007 [ arxiv : astro - ph/0210617 ] .
d. hooper , i. de la calle perez , j. silk , f. ferrer and s. sarkar , jcap * 0409 * ( 2004 ) 002 [ arxiv : astro - ph/0404205 ] .
n. fornengo , l. pieri and s. scopel , phys .
d * 70 * ( 2004 ) 103529 [ arxiv : hep - ph/0407342 ] .
d. horns , phys .
b * 607 * ( 2005 ) 225 [ erratum - ibid .
b * 611 * ( 2005 ) 297 ] [ arxiv : astro - ph/0408192 ] .
l. bergstrom , t. bringmann , m. eriksson and m. gustafsson , phys .
* 94 * ( 2005 ) 131301 [ arxiv : astro - ph/0410359 ] .
s. profumo , phys .
d * 72 * ( 2005 ) 103521 [ arxiv : astro - ph/0508628 ] .
g. zaharijas and d. hooper , phys .
d * 73 * ( 2006 ) 103501 [ arxiv : astro - ph/0603540 ] .
f. aharonian _ et al .
_ [ h.e.s.s .
collaboration ] , phys .
* 97 * ( 2006 ) 221102 [ erratum - ibid .
* 97 * ( 2006 ) 249901 ] [ arxiv : astro - ph/0610509 ] .
s. dodelson , d. hooper and p. d. serpico , arxiv:0711.4621 [ astro - ph ] .
p. gondolo , phys .
b * 494 * ( 2000 ) 181 [ arxiv : hep - ph/0002226 ] .
g. bertone , g. sigl and j. silk , mon . not .
* 326 * ( 2001 ) 799 [ arxiv : astro - ph/0101134 ] .
r. aloisio , p. blasi and a. v. olinto , jcap * 0405 * ( 2004 ) 007 [ arxiv : astro - ph/0402588 ] .
l. bergstrom , m. fairbairn and l. pieri , phys .
d * 74 * ( 2006 ) 123515 [ arxiv : astro - ph/0607327 ] .
s. colafrancesco , s. profumo and p. ullio , astron .
astrophys .
* 455 * ( 2006 ) 21 [ arxiv : astro - ph/0507575 ] . s. colafrancesco , s. profumo and p. ullio , phys .
d * 75 * ( 2007 ) 023513 [ arxiv : astro - ph/0607073 ] .
s. colafrancesco and b. mele , astrophys .
j. * 562 * ( 2001 ) 24 [ arxiv : astro - ph/0008127 ] .
p. blasi , a. v. olinto and c. tyler , astropart .
phys . *
18 * ( 2003 ) 649 [ arxiv : astro - ph/0202049 ] . g. bertone , g. servant and g. sigl , phys .
d * 68 * ( 2003 ) 044008 [ arxiv : hep - ph/0211342 ] .
a. tasitsiomi , j. m. siegal - gaskins and a. v. olinto , astropart .
* 21 * ( 2004 ) 637 [ arxiv : astro - ph/0307375 ] .
e. a. baltz and l. wai , phys .
d * 70 * ( 2004 ) 023512 [ arxiv : astro - ph/0403528 ] .
d. hooper , d. p. finkbeiner and g. dobler , phys .
d * 76 * ( 2007 ) 083012 [ arxiv:0705.3655 [ astro - ph ] ] .
r. narayan , r. mahadevan , j. e. grindlay , r. g. popham and c. gammie , astrophys .
j. * 492 * ( 1998 ) 554 [ arxiv : astro - ph/9706112 ] .
r. d. davies , d. walsh and r. s. booth , mnras * 177 * ( 1976 ) 319 .
f. melia and h. falcke , ann .
astrophys . * 39 * , 309 ( 2001 ) [ arxiv : astro - ph/0106162 ] .
z. q. shen , k. y. lo , m. c. liang , p. t. p. ho and j. h. p. zhao , nature * 438 * ( 2005 ) 62 [ arxiv : astro - ph/0512515 ] .
haslam , c. g. t. , salter , c. j. , stoffel , h. , & wilson , w. e. 1982 , a&as , 47 , 1 t. n. larosa , c. l. brogan , s. n. shore , t. j. lazio , n. e. kassim and m. e. nord , astrophys .
j. * 626 * ( 2005 ) l23 [ arxiv : astro - ph/0505244 ] .
t. n. larosa , n. e. kassim , t. j. w. lazio and s. d. hyman , astrophys .
j. * 119 * ( 2000 ) 207 .
r. genzel _ et al .
_ , nature * 425 * , 934 ( 2003 ) [ arxiv : astro - ph/0310821 ]
. y. clenet _ et al .
_ , astron .
astrophys .
* 417 * ( 2004 ) l15 [ arxiv : astro - ph/0312108 ] .
f. k. baganoff _ et al .
_ , astrophys . j. * 591 * ( 2003 ) 891 [ arxiv : astro - ph/0102151 ] .
y. d. xu , r. narayan , e. quataert , f. yuan and f. k. baganoff , astrophys . j. * 640 * ( 2006 ) 319 [ arxiv : astro - ph/0511590 ] .
muno _ et al .
_ , astrophys .
j. * 613 * ( 2004 ) 326 [ arxiv : astro - ph/0402087 ] .
h. a. mayer - hasselwander _ et al .
_ , astron .
astrophys . *
335 * , 161 ( 1998 ) .
f. aharonian and a. neronov , astrophys .
j. * 619 * ( 2005 ) 306 [ arxiv : astro - ph/0408303 ] .
f. aharonian _ et al .
_ [ h.e.s.s .
collaboration ] , nature * 439 * ( 2006 ) 695 [ arxiv : astro - ph/0603021 ] .
j. albert _ et al .
_ [ magic collaboration ] , astrophys . j. * 638 * ( 2006 ) l101 [ arxiv : astro - ph/0512469 ] .
k. tsuchiya _ et al .
_ [ cangaroo - ii collaboration ] , astrophys . j. * 606 * ( 2004 ) l115 [ arxiv : astro - ph/0403592 ] .
k. kosack _ et al . _
[ the veritas collaboration ] , astrophys .
j. * 608 * ( 2004 ) l97 [ arxiv : astro - ph/0403422 ] . c. van eldik , o. bolz , i. braun , g. hermann , j. hinton and w. hofmann [ h.e.s.s .
collaboration ] , arxiv:0709.3729 [ astro - ph ] .
q. d. wang , f. j. lu and e. v. gotthelf , mon . not .
* 367 * ( 2006 ) 937 [ arxiv : astro - ph/0512643 ] . k. ferriere , w. gillard and p. jean , astron .
astrophys .
* 467 * ( 2007 ) 611 [ arxiv : astro - ph/0702532 ] .
t. a. porter , i. v. moskalenko , a. w. strong , e. orlando and l. bouchet , arxiv:0804.1774 [ astro - ph ] .
m. morris , arxiv : astro - ph/0701050 .
r. f. coker and f. melia , [ arxiv : astro - ph/9909411 ] .
a. w. strong , i. v. moskalenko and v. s. ptuskin , ann .
nucl . part .
* 57 * ( 2007 ) 285 [ arxiv : astro - ph/0701517 ] .
d. r. ballantyne , f. melia , s. liu and r. m. crocker , astrophys . j. * 657 * ( 2007 ) l13 [ arxiv : astro - ph/0701709 ] .
f. melia , ap .
j. * 387 * , l25-l28,1992 j. bland - hawthorn and m. cohen , astrophys . j. * 582 * ( 2003 ) 246 [ arxiv : astro - ph/0208553 ] . c. j. law , arxiv:0705.0114 [ astro - ph ] .
g. rybicki and a.p .
lightman , 1979 , radiative processes in astrophysics , john wiley & sons inc .
p. gondolo and j. silk , phys .
* 83 * ( 1999 ) 1719 , [ arxiv : astro - ph/9906391 ] .
p. ullio , h. zhao and m. kamionkowski , phys .
d * 64 * ( 2001 ) 043504 , [ arxiv : astro - ph/0101481 ] . g. bertone and d. merritt , phys . rev .
d * 72 * ( 2005 ) 103502 , [ arxiv : astro - ph/0501555 ] .
a. klypin , h. zhao and r. s. somerville , astrophys .
j. * 573 * ( 2002 ) 597 [ arxiv : astro - ph/0110390 ] .
f. prada , a. klypin , j. flix , m. martinez and e. simonneau , arxiv : astro - ph/0401512 . m. gustafsson , m. fairbairn and j. sommer - larsen , phys . rev .
d * 74 * , 123522 ( 2006 ) [ arxiv : astro - ph/0608634 ] .
m. regis , m. serone and p. ullio , jhep * 0703 * , 084 ( 2007 ) [ arxiv : hep - ph/0612286 ] .
r. morrison and d. mccammon , astrophys .
j. * 270 * ( 1983 ) 119 .
zhao , p. , jerius , d. h. , edgar , r. j. , gaetz , t. j. , van speybroeck , l. p. , biller , b. , beckerman , e. , & marshall , h. l. 2004 , proc .
spie , 5165 , 482
|
we perform a systematic study of the multi wavelength signal induced by weakly interacting massive particle ( wimp ) annihilations at the galactic center ( gc ) .
referring to a generic wimp dark matter ( dm ) scenario and depending on astrophysical inputs , we discuss spectral and angular features and sketch correlations among signals in the different energy bands .
none of the components which have been associated to the gc source sgr a@xmath0 , nor the diffuse emission components from the gc region , have spectral or angular features typical of a dm source .
still , data - sets at all energy bands , namely the radio , near infrared , x - ray and gamma - ray bands , contribute to place significant constraints on the wimp parameter space . in general , the gamma - ray energy range is not the one with the largest signal to background ratio . in the case of large magnetic fields close to the gc ,
x - ray data give the tightest bounds .
the emission in the radio - band , which is less model dependent , is very constraining as well .
the recent detection by hess of a gc gamma - ray source , and of a diffuse gamma - ray component , limits the possibility of a dm discovery with the next generation of gamma - ray telescopes , like glast and cta .
we find that most of the region in the parameter space accessible to these instruments is actually already excluded at other wavelengths . on the other hand , there may be still an open window to improve constraints with wide - field radio observations .
|
transport properties of low - dimensional systems have been subject of intensive research work over the last two decades .
one of the fundamental discoveries that has driven the field was the observation of conductance quantization in ballistic quantum wires and quantum point contacts . @xcite it was found that conductance exhibits a staircase - like dependence on the electron density with the universal step .
the understanding of this phenomenon follows already from the single - electron picture , which predicts for the conductance of a one - dimensional single channel - clean wire , @xcite @xmath0 the physical origin of conductance plateaus at certain gate voltages was associated with a fixed number of occupied electronic subbands , each supplying one quantum of conductance @xmath1 . within the same approach of noninteracting particles both charge and energy
are carried by electronic excitations .
this results in the universal relation between electric and thermal conductances , known as the wiedemann - franz law @xmath2 .
the thermal conductance of noninteracting electrons is thus @xmath3 in addition to @xmath4 and @xmath5 two thermoelectric coefficients of the electron gas are usually of great interest
. these are thermopower @xmath6 , which relates an induced voltage drop across the wire to applied temperature gradient , and peltier coefficient @xmath7 connecting electric and heat currents .
these two coefficients are connected by an onsager relation @xmath8 . in the absence of interactions
the thermopower and peltier coefficients are exponentially small @xmath9 at low temperatures @xmath10 .
( here @xmath11 is the chemical potential . )
the reason for such strong suppression of thermopower and peltier coefficients is the partial cancellation between heat currents carried by particles with energies @xmath12 and @xmath13 .
only the absence of electronic states below the bottom of the band prevents @xmath7 and @xmath6 from vanishing exactly .
the remarkable success of the simple single - electron picture in describing the quantization of conductance and explaining the temperature dependence of thermoelectric coefficients is attributed to the fact that quantum wires are always connected to two - dimensional leads , where interactions between electrons do not play a significant role . even though the interactions in the wire are usually not weak , i.e. , @xmath14 , where @xmath15 is the fermi velocity , it has been shown within the so - called luttinger - liquid model of one - dimensional electrons that the interactions inside the wire do not affect conductance quantization .
@xcite it is no surprise then that a number of recent experiments , @xcite that revealed deviations from the perfect quantization , , in low - density wires , attracted a great deal of theoretical attention .
@xcite these deviations often take the form of a shoulder - like feature , which develops at finite temperature just below the first quantized plateau , around @xmath16 . at present
there is no consensus on the theoretical interpretation of this phenomenon .
however , it is generally accepted that electron - electron interaction effects should be involved in explaining these experimental observations . in a number of recent publications @xcite transport properties of one - dimensional conductors
were reconsidered focusing on the physics which lies beyond an ideal luttinger - liquid model . in particular , when studying the temperature dependence of the corresponding kinetic coefficients refs
. emphasized one fundamental aspect of interactions , namely the role of physical processes that lead to equilibration of electrons inside the wire .
it should be emphasized that equilibration is absent in an ideal luttinger liquid since bosonic elementary excitations of the latter have infinite lifetime , thus there is no relaxation towards equilibrium in these systems , no matter how strong the interactions are . in higher - dimensional systems equilibration at low temperatures is primarily provided by pair collisions of electrons .
these , however , do not provide relaxation in one - dimensional systems .
this is due to the conservation laws for momentum and energy which severely restrict the phase space available for scattering . as a result
, pair collisions in ideal one - dimensional wires can occur with a zero momentum change or an interchange of the two momenta , leaving the distribution function unaffected .
the leading equilibration mechanism thus involves collisions of more than two particles . for a weakly interacting system , it is then natural to assume that equilibration is provided by three - particle scattering processes .
@xcite this , of course , also relies on the additional assumption that other degrees of freedom , which can absorb energy and momentum from electrons ( phonons , for example ) can be ignored .
this assumption is acceptable in many cases since electron - phonon coupling constant is typically much smaller than that due to the electron - electron interactions . in practice ,
long one - dimensional structures are strongly prone to inhomogeneities inevitably present due to the nearby gates or charged dopants underlying the wire . however , most preceding works studied effect of equilibration on transport assuming uniform ( clean ) wires .
the notable exceptions include refs . and where smooth inhomogeneities were accounted for while assuming full equilibration of the electronic system .
the purpose of the present work is to study effects of inhomogeneities on transport properties of partially equilibrated quantum wires .
we focus our attention to the situation where the scale of inhomogeneities @xmath17 is much larger than the electron fermi wavelength , @xmath18 , such that backscattering of electrons from the inhomogeneities is negligible . in this case
only interactions ( three - particle collisions ) may interrupt direct flow of the electron liquid and convert ( backscatter ) some right - moving electrons into left - moving ones .
since non - uniform systems are no longer translationally invariant , there is an additional scattering mechanism , which can relax electron momentum without changing the number of right- and left - moving particles .
@xcite we find that equilibration due to three - particle collisions and inhomogeneity - induced momentum - nonconserving scattering compete with each other .
an interplay between these two effects leads to a very interesting picture of the electronic transport and results in temperature - dependent corrections to the wire resistance and thermoelectric coefficients .
the paper is organized as follows . in sec .
[ sec - unifromwire ] we discuss the general structure of the electron distribution function and transport in a clean one - dimensional wire .
we briefly mention the resulting transport coefficients for this case , which were recently reported in ref . .
in sec . [ sec - formalism ] we develop a general formalism which enables us to treat equilibration of the electron system due to three - particle collisions and inhomogeneity - induced scattering on equal footing .
the central part of our work is sec .
[ sec - transport ] where we apply this theory to the calculation of transport coefficients in one - dimensional inhomogeneous wires . from our general expressions
we recover the results known for the short and long uniform wires , and also consider several experimentally relevant simple models of inhomogeneous case .
we summarize our results in sec .
[ sec - discussions ] .
supplementary appendices accompany some technical aspects of our calculations .
in the absence of interactions , left- and right - moving electrons inside the wire are at equilibrium with the reservoir they originated from .
if a voltage bias @xmath19 and/or temperature difference @xmath20 is applied between the reservoirs , then corresponding equilibria differ from each other , giving rise to a particular form of the nonequilibrium distribution function inside the wire .
this distribution depends on the direction of motion of electrons and for the right- and left - moving particles is controlled , respectively , by the left and right lead , @xmath21 here @xmath22 is the energy of an electron with momentum @xmath23 and @xmath24 is the unit step function .
the difference between the chemical potentials ( temperatures ) in the leads is equal to the voltage ( temperature difference ) applied to the wire @xmath25 ( @xmath26 ) . using the distribution function at @xmath27
one can find electric current @xmath28 with the conductance of noninteracting electrons @xmath29 , thus recovering , eq .
, up to an exponentially small correction . the same distribution , eq . , provides thermal conductance , eq . , and thermoelectric coefficients @xmath30 , consistent with eq . .
in the presence of interactions ballistic propagation of electrons through the wire may be interrupted by collisions with other electrons . as a result of these collisions ,
some electrons change their direction of motion thus losing memory of the lead they originated from .
such backscattering processes modify the electron distribution function which is then no longer given by eq . .
it is important to realize that the effect of electron collisions on the distribution function depends strongly on the length of the wire .
indeed , electrons traverse short wires relatively fast , such that interactions do not have time to change distribution , eq . , considerably . on the other hand , in the limit of very long wire one should expect full equilibration of left- and right - moving electrons into a single distribution , even in the case of weak interactions . for the galilean invariant system one can easily infer the electron distribution function in a fully equilibrated state .
indeed , viewed from a reference frame moving with the drift velocity @xmath31 ( where @xmath32 is the electric current and @xmath33 is the electron density ) the electron system is at rest and must be described by the equilibrium fermi distribution . performing a galilean transformation back into the stationary frame of reference this distribution takes the form , @xmath34 where the chemical potential @xmath35 and temperature @xmath36 inside the equilibrated wire are , in general , different from @xmath37 and @xmath38 . at zero temperature , @xmath39 , the distributions , eqs . and ,
coincide , provided @xmath40 , where @xmath41 is the fermi momentum of the system . at non - zero temperature the distribution function , eq .
, of electrons inside the equilibrated wire is slightly different from the distribution , eq . , supplied by the leads . the mismatch between the two distribution functions results in additional resistance , reducing the conductance of noninteracting electrons to @xcite @xmath42\,.\ ]
] this result is universal since it was obtained without making any specific assumptions regarding the process of equilibration .
the corresponding derivation relied uniquely on the analysis of conservation laws for energy , momentum , and particle number .
it is applicable as long as wire length @xmath43 exceeds certain equilibration length @xmath44 such that the distribution , eq .
, is already established . the exact definition of @xmath44 is model specific and depends on the interaction between electrons .
quite generically , however , it can be argued that this length is exponentially large at low temperature @xmath45 .
the exponential scale can be understood from the mechanism of equilibration @xcite which is also discussed later in the text .
the thermal conductance of fully equilibrated wire is zero @xmath46 this result can be understood simply from the structure of the distribution function .
one should recall that thermal conductance is defined under the condition that electric current @xmath47 vanishes .
thus @xmath48 and the distribution , eq . , takes the form of the standard fermi - dirac distribution .
due to its symmetry @xmath49 the heat current carried by electrons vanishes , regardless of the temperature bias @xmath20 applied to the wire .
one therefore finds that in an infinitely long wire the thermal conductance is zero . one should note , however , that thermal conductivity is finite,@xcite i.e. , at @xmath50 the thermal conductance scales as @xmath51 . unlike the electric and thermal conductances , thermopower and peltier coefficients
are significantly enhanced by equilibration effects .
specifically , @xmath7 grows from the exponentially small value , eq .
, for short wires @xmath52 to @xmath53 in fully equilibrated ( long ) wires , @xmath54 .
a more careful treatment of the equilibration effects is required in wires of intermediate length @xmath55 , where the electron distribution function is only partially equilibrated .
@xcite as we already mentioned , in the case of weakly interacting electrons the leading mechanism of equilibration is provided by three - particle collisions . at low temperatures
one should consider two types of such collisions .
the strongest scattering events involve three particles near the fermi level ( for example , one left mover that scatters off two right movers such that all particles preserve their direction of motion ) .
these collisions are relatively fast , and the corresponding scattering length @xmath56 scales as a power law of temperature .
however , these collisions alone can not establish the distribution since they conserve the number of right- and left - moving particles .
the other important three - particle collisions involve backscattering of , say , a right - moving electron into a left - moving one .
this backscattering occurs near the bottom of the band and provides equilibration between the chemical potentials of right and left movers , thus establishing the distribution , eq . .
since at low temperatures the probability to find an empty state at the band bottom is exponentially small , the corresponding relaxation process is very slow , and equilibration length is large , @xmath45 .
let us consider now a segment of the wire , whose length @xmath57 is small compared to the equilibration length @xmath44 but large as compared to @xmath56 , namely , @xmath58 .
this condition implies that typical electron with energy near the fermi level passes through the segment without backscattering so that the distribution , eq .
, can not be established .
on the other hand , @xmath57 is already sufficiently large for electrons to experience other multiple collisions which allow momentum and energy exchange between right- and left - moving electrons . under these conditions ,
the electron distribution function in the segment achieves a state of partial equilibration , in which the numbers @xmath59 and @xmath60 of the right- and left - moving electrons are conserved independently .
the form of this distribution can be obtained from the general statistical mechanics argument by maximizing the entropy of electrons while preserving @xmath61 , total energy and momentum of the system , @xcite @xmath62 here @xmath63 is the effective temperature , parameter @xmath64 has dimension of velocity and accounts for the conservation of momentum in electron collisions , and @xmath65 are the chemical potentials of the left- and right - moving particles .
the distribution , eq . ,
smoothly interpolates between the regimes of no equilibration , eq . , and that of full equilibration , eq . . in the absence of temperature difference ,
@xmath66 , the unperturbed distribution , eq . , is obtained from eq . by setting @xmath67 and identifying the chemical potentials with those in the leads : @xmath68 and @xmath69 .
[ here and throughout the paper we use @xmath70 to denote left ( right ) lead while @xmath71 denote left ( right ) movers . ] the fully equilibrated distribution , eq .
, is obtained from eq . by setting @xmath72 . in this case
the electric current is expressed as @xmath73 , which identifies parameter @xmath64 as the drift velocity @xmath74 .
we should emphasize here that since the distribution , eq .
, is applicable for the segment of the wire outlined above then all four parameters @xmath75 and @xmath76 defining @xmath77 are , in principle , coordinate dependent .
the implications of the distribution , eq . , for the transport coefficients of partially equilibrated clean wires were discussed in ref . .
in the following we generalize the above picture of electronic transport in one - dimensional wires accounting for possible non - uniformities of the system .
inhomogeneous quantum wire connected adiabatically to two - dimensional leads and biased by the small voltage @xmath25 and/or temperature difference @xmath26 .
we assume that the spatial scale @xmath17 of the inhomogeneities is large compared to the fermi wavelength , and thus electrons do not experience backscattering from inhomogeneities .
in this case only three - particle equilibration processes may interrupt the flow of right - moving electrons and convert some right movers into left movers .
wavy lines represent heat exchange @xmath78 between right movers and left movers.,width=302 ]
consider an inhomogeneous quantum wire of length @xmath43 , connected by ideal reflectionless contacts to noninteracting leads and biased by a small voltage @xmath19 and/or temperature difference @xmath20 , see fig .
[ fig - geometry ] .
if the spatial variations related to inhomogeneities occur on a length scale @xmath17 much larger than the fermi wavelength @xmath79 , electrons do not suffer any backscattering .
since the physical picture of equilibration in one - dimensional wire can be readily understood at the level of weakly interacting electrons we restrict our attention to this case and describe the system in the framework of kinetic equation . in this case
electron distribution function @xmath80 obeys the boltzmann equation , @xmath81 where static potential @xmath82 accounts for inhomogeneities of the wire and @xmath83 conventionally stands for the collision integral .
we are interested in the steady - state regime when distribution function @xmath80 does not depend explicitly on time , and thus set @xmath84
. it will be also convenient to split the distribution @xmath85 into two parts corresponding to the right and left movers , @xmath86 and express it as the function of energy @xmath87 for the given momentum @xmath23 .
kinetic eq . should be supplemented by the boundary conditions at the ends of the wire that are controlled by the leads , [ f - rl - boundary ] @xmath88 @xmath89 where @xmath90 , @xmath91 and @xmath92 , @xmath93 .
[ we use shorthand notation for the distribution function of right movers at the left lead @xmath94 and left movers at the right lead @xmath95 . ] the parametrization , eq . , is especially useful since owing to the simple algebraic relation , @xmath96 the inhomogeneity - related term drops out from the left - hand side of the kinetic equation , except for the residual contribution @xmath97 $ ] at @xmath98 . for noninteracting electrons
the mismatch between distribution functions @xmath99 of right and left movers is exponentially small at the bottom of the band .
in addition , even this small discontinuity is smeared by inter - electron scattering responsible for equilibration.@xcite it is thus safe to take @xmath100 for @xmath98 and we get then instead of eq .
, @xmath101 as the first step of our general analysis , we demonstrate now with the help of kinetic eq . that deviations in electric and thermal conductances from their noninteracting values [ eqs . and ]
are ultimately related to the rate of change in the number of say right - moving electrons @xmath102 and heat exchange rate @xmath78 between right movers and left movers .
the rate of change in the number of right movers @xmath102 due to electron collisions is obtained from the collision integral @xmath83 upon integration over positive momenta and wire length , @xmath103 where the coefficient 2 stands for two spin projections .
owing to the boltzmann eq .
@xmath102 can be equivalently presented in terms of the distribution function of right - moving electrons as @xmath104 we can integrate this expression by parts by noticing that @xmath105 and approximating in the following the distribution function of right movers by unity at the bottom of the band , which is correct up to exponentially small terms at low temperatures @xmath106 $ ] .
this would give then @xmath107\,,\ ] ] where we used standard definition for the currents of right / left movers , @xmath108 in eq .
the incoming current @xmath109 of right movers at @xmath110 is known since it is controlled by the distribution of noninteracting electrons in the left lead , eq . . however , the outgoing @xmath111 is not known because the distribution function of right - movers varies along the wire as a result of scattering .
it is convenient to exclude this unknown from eq . by noticing that the total current @xmath112 in the wire does not depend on position due to the conservation of the number of electrons and
can be written as @xmath113 .
this results in @xmath114+\frac{2}{h}[u(r)-u(l)]\,.\ ] ] the benefit of performing this step is that now both currents @xmath115 and @xmath116 are controlled by the noninteracting leads whose distribution functions are given by the boundary conditions [ eq . ] . furthermore ,
since @xmath117 and @xmath112 vanish in the absence of applied bias , we can exclude the @xmath118-dependent contribution from eq . by subtracting from it
@xmath119 $ ] .
this leads to @xmath120\,.\ ] ] the difference between currents of right movers at the left boundary with and without bias in eq .
can be found with the help of distribution function . indeed , after a simple calculation @xmath121=\frac{2ev}{h},\nonumber\end{aligned}\ ] ] valid up to corrections small as @xmath122 , we find @xmath123 where @xmath124 .
this result can be thought of as a generalization of landauer formula for interacting one - dimensional systems . in the noninteracting limit @xmath125 and we recover @xmath126 while a finite @xmath102 would lead to a change in the conductance .
equation was derived earlier for uniform ( clean ) wires .
@xcite we have shown here that it remains intact even in the case of inhomogeneous wires .
we now repeat the above calculation for the energy change of right - movers @xmath127 induced by electron collisions .
the latter is obtained from the collision integral @xmath83 by multiplying it by @xmath128 and then integrating over positive momenta and the wire length @xmath129 it can be equivalently rewritten in terms of @xmath130 by making use of the boltzmann eq .
, @xmath131 after integration by parts , similar to eq . , one finds @xmath132\,,\ ] ] where we used the usual definition for energy currents of right / left movers , @xmath133 conservation of energy ensures that the energy current @xmath134 is constant along the wire . by using it at the right end @xmath135
, we can exclude unknown @xmath136 from eq . in analogy with eq . .
in addition , since @xmath137 and @xmath138 without the bias , we subtract @xmath139 $ ] to exclude the @xmath118-dependent contribution .
this procedure gives for the rate of energy change @xmath140\,.\ ] ] the energy current of right movers at the left end of the wire is controlled by the noninteracting lead with known distribution function .
to linear order in @xmath19 and @xmath141 a simple calculation gives us @xmath142\nonumber \\ & & = \frac{2ev\mu}{h}+\frac{2\pi^2t\delta t}{3h}\,.\end{aligned}\ ] ] it is convenient to combine electric and energy currents into the heat current @xmath143 for which we find from eqs . and , @xmath144 where @xmath145 is the heat transferred into the right - moving subsystem by electron collisions . as expected when @xmath146 we recover from eq .
the thermal conductance of noninteracting electrons eq . , while a nonvanishing @xmath78 results in an interaction - induced change in @xmath5 .
equation was reported earlier for uniform wires @xcite and as it is shown here remains valid even in the inhomogeneous case .
so far our basic eqs . and contain four unknown entries : two response currents @xmath32 and @xmath147 due to applied bias @xmath19 and temperature difference @xmath20 , and two rates @xmath102 and @xmath78 nonvanishing due to interactions . in the next sections we demonstrate that all four quantities can be expressed in terms of only two parameters @xmath148 and @xmath149 which enter the distribution function of partially equilibrated electrons .
this would give us the closed set of equations that relate @xmath150 and thus determine the transport coefficients of interest .
electric and heat currents can be easily found knowing the distribution function @xmath85 of electrons in the wire . as we discussed in sec .
[ sec - unifromwire ] for the partially equilibrated wire this distribution is given by eq . .
it is worth emphasizing that for the inhomogeneous case all four parameters entering eq . :
velocity @xmath64 , chemical potential of right- and left - moving electrons @xmath151 , and effective temperature @xmath63 , are , in principle , coordinate dependent .
furthermore , the distribution , eq . , does not apply to particles near the bottom of the band , for @xmath152 , as explained later in the text ( see also corresponding discussions in ref . ) .
this , however , does not cause any extra difficulties since transport quantities of interest are determined by the behavior of the distribution function near the fermi level . by using the definition of the current , eq . , and distribution function we obtain the electric current in the partially equilibrated wire @xmath153 notice here that although contributions to the current due to the electron drift , @xmath154 , and partial equilibration between right and left movers , @xmath155 , are individually coordinate dependent , their sum must be constant along the wire .
this is a consequence of the particle number conservation .
since the heat current @xmath147 also does not depend on position , it can be calculated at any point in the wire . in the regions not too close to the leads the distribution function
is expected to have the partially equilibrated form . then using expressions and @xmath156 for @xmath112 and @xmath134 we obtain after sommerfeld expansion of the integrand to leading order in @xmath157 , @xmath158 where we introduced @xmath159 .
since @xmath147 is already proportional to small @xmath160 , we were able to replace @xmath63 with @xmath161 within the linear - response regime .
notice also that the particular combination @xmath162 defining heat current @xmath147 must be coordinate independent . to make further progress we should elaborate on the expressions for the rates @xmath102 and @xmath78 ,
whose explicit forms depend on details of the equilibration mechanism . this can be done following the idea suggested in ref . and
we show below how @xmath102 and @xmath78 can be expressed through @xmath163 and @xmath148 .
( a ) dominant three - particle collision which changes the number of right - moving electrons .
( b ) equilibration mechanism , multi - step backscattering of right mover into the left mover . at low temperatures
each step @xmath164 in momentum space is of the order of @xmath165 .
( c ) energy - conserving two - particle scattering process that violates conservation of momentum .
this process is possible due to the presence of inhomogeneities.,width=302 ] first let us identify the leading backscattering mechanism that contributes to @xmath102 .
the most favorable collisions should involve a maximal number of electronic states close to the fermi points .
however , due to the conservation of total energy and momentum , collisions that change the number of right and left movers can not occur near the fermi level only , and have to involve states deep in the electron band . as was pointed out in ref . the scattering process most important in altering the current thus typically scatters two electrons close to the fermi points and one electron at the bottom of the band , as schematically depicted in fig .
[ fig-3pc]a .
it is convenient to think of this collision as a process in which a deep hole , corresponding to the outgoing electron state , is backscattered by electron excitations close to the fermi level . these excitations are typically associated with a momentum change @xmath166 due to fermi blocking .
let us furthermore characterize this process by introducing three - particle scattering rate @xmath167 , which can be approximated by a constant because the initial and final states both lie at the bottom of the band . since the sign of @xmath164 varies in a random fashion from one collision to another the hole performs a brownian motion in momentum space .
the corresponding diffusion coefficient @xmath168 can be readily estimated . the typical momentum change of a hole over time @xmath169 behaves as @xmath170 .
as we assumed the hole changes its momentum by @xmath171 once during the time @xmath172 , so we conclude that @xmath173 for @xmath174 and thus estimate @xmath175 the change @xmath176 in the number of right - moving electrons over the time @xmath177 for the segment of the wire @xmath178 is given by the rate @xmath179 times the number of deep holes susceptible to be backscattered .
the latter can be estimated from the probability to find a left- or right - moving hole @xmath180 and the number of states @xmath181 available within the typical momentum range @xmath182 of the backscattering processes shown in fig .
[ fig-3pc]b . taking into account that the scattering of left- and right - moving holes both contribute to @xmath102 , but with opposite signs
, one finally estimates @xmath183 with @xmath184 .
a careful calculation based on the kinetic equation gives @xcite @xmath185 the expression for the diffusion coefficient @xmath168 is model specific .
our preliminary estimate gives @xmath186 in the case of coulomb interaction .
we continue now with the calculation of the rate @xmath78 , which consists of two contributions @xmath187 the first one @xmath188 is related to the same backscattering events that change the number of right - movers @xmath117 .
exploring the fact that both rates are caused by the same physical processes it was shown in refs . and that there is a relation between @xmath102 and @xmath189 , which we generalize here for the inhomogeneous case , @xmath190 the logic behind this equation is as follows .
the backscattering processes transform the unperturbed distribution of electrons into the partially equilibrated form with nonvanishing @xmath148 .
the two distributions differ most prominently at energies within @xmath191 of the fermi level .
one can thus assume that all the right - moving electrons contributing to @xmath102 are removed from the vicinity of the right fermi point and placed to the vicinity of the left one .
each such transfer reduces the momentum of the system by @xmath192 .
the other electrons have to be scattered in the vicinities of the two fermi points to accommodate this momentum change . in the special case of three - particle collisions ,
the transfer of electron from the right fermi point to the left one is accomplished in a number of small steps with momentum change @xmath193 , and at each step one additional electron is scattered near each of the two fermi points , see fig .
[ fig-3pc]b . as a result of the rearrangement of electrons near the two fermi points
, the local momentum change @xmath192 of the backscattered electrons is distributed between the remaining right- and left - moving electrons , i.e. , @xmath194 .
thus the energy of the remaining right movers increases by @xmath195 whereas that of the left movers decreases , @xmath196 .
then , the conservation of energy requires @xmath197 . in the end , the energy balance for the right - moving electrons consists of a loss of @xmath11 due to the removal of one particle from the fermi level and a gain of @xmath198 due to the redistribution of momentum . as a result , for every right - moving electron that changes direction , @xmath199 , the right - movers energy increases by an amount @xmath200 ,
so one concludes that @xmath201 or equivalently @xmath202 .
equation follows from here naturally if one applies the same argument but for the segment of wire @xmath178 such that rates @xmath102 and @xmath78 are accounted per unit of length in the inhomogeneous wire .
the other contribution @xmath203 to the heat transferred by right movers in eq .
is due to scattering processes that do not conserve momentum , see fig .
[ fig-3pc]c for illustration .
these two - body collisions are possible only in the inhomogeneous case .
they do not change the number of right - moving electrons , but do change their energy .
it is expected that this rate is proportional to the velocity @xmath64 of the electron liquid , @xmath204 .
indeed , two - particle collisions of fig .
[ fig-3pc]c involve a right mover with momentum @xmath205 and a left mover with momentum @xmath206 . for these electrons the drift term @xmath207 in the distribution function
can be absorbed into the temperature @xmath63 , such that right movers can be considered as being at an effective temperature @xmath208 while left movers at temperature @xmath209 , to linear order in @xmath64 .
@xcite according to the general principle of statistical mechanics thermalization between these subsystems involves the energy flow from `` warmer '' right movers to `` colder '' left movers that is proportional to the difference in temperatures between the two , @xmath210 .
an explicit microscopic calculation of the rate @xmath211 done in appendix ( [ sec - appendix - qr ] ) gives @xmath212 here @xmath213 is a scattering length scale associated with these momentum - nonconserving collisions , fig .
[ fig-3pc]c , @xmath214 where the parameter @xmath215 ^ 2\right.\nonumber\\ & & \hskip-.75cm\left.+\left[\partial_{x}\ ! \left(\frac{\mathcal{v}_0}{\pi\hbar v_f(x)}\right)\right]^2\!\!\
! + \left[\partial_{x}\ ! \left(\frac{\mathcal{v}_{2k_f(x)}}{\pi\hbar
v_f(x)}\right)\right]^2 \right\}\end{aligned}\ ] ] is expressed through the zero momentum and @xmath216 fourier components of the electronic interaction potential @xmath217 .
the complete rate @xmath78 is thus given by the sum of eqs . and and we find @xmath218 one should make two important comments regarding this result .
first , the effect of inhomogeneity on resistivity of a quantum wire was recently addressed in refs . assuming that electrons are fully equilibrated and thus described by the distribution function .
this assumption requires that three - body interaction processes which change the number of right - moving electrons dominate over the momentum - nonconserving scattering .
however , in a situation where both interaction processes happen on a comparable time scale the system is frustrated with a finite value of @xmath219 and thus @xmath220 , so that electrons are described by the distribution , eq . , which we used in our calculations .
to elucidate further the origin of the frustration it is important to emphasize that momentum - nonconserving scattering resists full equilibration of electrons into a single distribution , eq . .
these scattering processes reduce velocity @xmath64 and thus prevent complete relaxation of the difference in chemical potentials @xmath221 . indeed , since the current @xmath32 is fixed by the external circuit then according to eq .
decrease in @xmath64 implies increase in @xmath221 .
this effect is opposite to that of the equilibration processes due to three - particle collisions which tend to relax @xmath221 .
second , the correction to resistance of a nonuniform wire was obtained in studies @xcite by calculating the rate of momentum change @xmath222 for right - moving electrons .
note , however , that in an inhomogeneous system without translational invariance momentum is not a good quantum number .
indeed , the momentum change of right movers due to collisions depends on position and thus , is not the same for electrons inside the wire , where it was calculated , @xcite than that ( actual change ) in the leads .
our present scheme is free from this difficulty . as we show in the next section it is really the rate @xmath78 , computed from the well - defined energy exchange , that is needed to determine the wire resistance and other transport coefficients .
we are set now for the calculation of transport coefficients in an inhomogeneous wire . indeed , our basic eqs .
and provide electric @xmath32 and heat @xmath147 currents as the response to the applied bias @xmath19 and temperature difference @xmath20 .
interaction effects are captured by the rates @xmath102 and @xmath78 induced by particle collisions that encode change in the number of right - moving electrons and energy exchange between right- and left - movers , respectively .
these rates are defined by eqs . and @xmath223 which still contain the unknown difference between the chemical potentials of partially equilibrated right and left movers @xmath163 and flow velocity @xmath148 .
equations and are the final ingredients that allow to establish the correspondence @xmath224 and thus find transport coefficients of interest .
technically one proceeds as follows .
first , employing conservation of currents ( recall that @xmath32 and @xmath147 are constant along the wire ) one can express @xmath163 and @xmath148 in terms of @xmath32 and @xmath147 from eqs . and .
second , one brings these relations into eqs . and @xmath223 to find the rates @xmath102 and @xmath78 in terms of @xmath32 and @xmath147 , which is possible to do in quadratures .
finally , eqs . and
define the desired correspondence @xmath224 . employing this procedure we find two independent linear equations , @xmath225- 2ir_1\overline{\mu}\,,\label{eq2}\end{aligned}\ ] ] where we introduced the dimensionless parameter , @xmath226
which quantifies the rate of three - particle processes that change number of right - moving electrons [ see eq .
] , as well as the weighted chemical potentials along the wire , @xmath227 from eqs . and
we find the resistance @xmath228\,,\ ] ] where @xmath229 and peltier coefficient @xmath230 in addition we find the thermal conductance @xmath231 and thermopower @xmath232 predictably , the peltier coefficient , eq . , and thermopower , eq . ,
satisfy the onsager relation @xmath8 .
are the main results of this paper . in the following we analyze these general expressions for a few modeling examples of inhomogeneities .
first of all , we perform a consistency check for the case of a uniform wire , recently studied in ref . .
in the homogeneous case @xmath233 all quantities defining @xmath234 , @xmath7 , @xmath5 and @xmath6 become coordinate independent : @xmath235 so that @xmath236 and @xmath237 . at the same time
@xmath238 which is a consequence of momentum conservation : in a uniform wire two - electron scattering processes , shown in fig .
[ fig-3pc]c , are not allowed . as a result ,
interaction - induced correction to the wire resistance , eq . , reduces to @xmath239 where @xmath240 . in order to establish a connection with the notations of ref .
we invert resistance , eq . , with @xmath241 taken from eq . , to get the conductance @xmath242 to leading order in @xmath157 and find @xmath243\,,\ ] ] where following ref .
we have introduced the equilibration length @xmath244 this result shows that for a long wire @xmath54 the conductance saturates to its length independent value , which still exhibits noticeable power - law correction in temperature , @xmath245 , already mentioned in sec .
[ sec - unifromwire ] [ eq . ] .
this saturation of conductance is expected , since the electronic system reaches full equilibrium . for short wires , @xmath246 ,
the interaction - induced correction to conductance is exponentially small @xmath247 and scales linearly with the length of the wire .
it is worth noting that eq .
is only applicable to wires longer than @xmath248 .
this constraint comes from the approximations made when deriving the rate of change for the right - moving electrons @xmath102 in eq .
[ see ref . for details ] . the peltier coefficient and thermopower of a uniform wire follow from eqs . and , and read @xmath249 it shows that @xmath7 grows from exponentially small values at @xmath52 to @xmath250 quoted in eq .
at @xmath54 .
the thermal conductance behaves very differently though .
one finds from eq . in the uniform limit @xmath251 at @xmath52 one recovers the result , eq . , for noninteracting wires , but as the length of the wire grows , @xmath5 is suppressed as @xmath252 and vanishes for fully equilibrated wires , see our earlier discussion presented below eq . .
equations and recover the corresponding results of ref . .
as the simplest prototype of nonuniform system we study two uniform wires of lengths @xmath253 and @xmath254 , with different densities , connected in series with each other .
we ignore here small contribution to transport coefficients coming from @xmath255 , which is nonzero only in the near vicinity of the junction between the wires and vanishes everywhere else . applying then eqs .
to this setup we find for the interaction - induced resistance , eq . , @xmath256 ^ 2 \frac{l_{1}l_{2}}{\ell^{(1)}_{\mathrm{eq}}\ell^{(2)}_{\mathrm{eq } } } } { 1+\frac{l_1}{\ell^{(1)}_{\mathrm{eq}}}+\frac{l_2}{\ell^{(2)}_{\mathrm{eq}}}}\,,\ ] ] with chemical potentials @xmath257 and equilibration lengths @xmath258 within each wire @xmath259 .
it is of special interest to consider the limit when one wire is infinitely long . upon taking the limit
@xmath260 the expression for @xmath241 reduces to @xmath261 ^ 2\frac{l_2}{\ell^{(2)}_{\mathrm{eq}}}\,.\ ] ] the first term of this formula corresponds to the residual resistance of the first wire .
it is independent of its length @xmath253 , which is natural since in the limit @xmath260 electrons reach full equilibration and the interaction - induced resistance should saturate .
@xcite the other contribution to the resistance in eq .
is due to the second wire , which , however , ceases to saturate even when @xmath262 .
thus it remains proportional to the wire length @xmath254 and could be much larger than the first term .
the absence of equilibration in the second wire is rather counterintuitive .
this result forced us to reexamine more carefully continuity equations for electric and heat currents .
we have found that it is not possible to match both @xmath32 and @xmath147 in the wires while simultaneously imposing vanishing chemical potential difference @xmath221 between right- and left - moving electrons .
indeed , let us suppose that in the limit @xmath263 and @xmath262 both wires are fully equilibrated , so that @xmath264 in each wire .
the current conservation then becomes @xmath265 , where @xmath266 is the drift velocity within each wire . according to eq .
the heat current in this case can be written as @xmath267 and clearly @xmath268 since @xmath269 .
the resolution of this controversy is possible only if @xmath219 at least within one wire even though its length exceeds the corresponding equilibration length . in appendix [ sec - appendix - r ]
we rederived eq .
relying on conservation laws only .
other transport coefficients do not show dramatic changes compared to a single uniform wire and their behavior follows expectedly as a natural generalization of eqs . and .
we find for the peltier coefficient of two connected uniform wires @xmath270 @xmath7 saturates to @xmath271 depending on which wire is fully equilibrated .
the thermal conductance is found to be @xmath272 which is natural generalization of eq . .
we now study more generic models of a nonuniform wire .
we assume only that disorder variations happen on the large spatial scale , @xmath273 , and concentrate on the case @xmath274 . in this case
the @xmath275 term in eq . can be ignored , and the interaction - induced resistance of the wire @xmath241 can be written as @xmath276 where we introduced @xmath277 .
expression is applicable to any realization of long - range disorder potential .
we now apply it to two special cases which allow simple analytical solution .
first , let us assume that amplitude @xmath278 of variations in the inhomogeneity potential along the wire is small , @xmath279 .
it turns out that eq .
covers three distinct regimes depending on the temperature . at lowest temperatures @xmath280 , where @xmath281}\,,\ ] ] three - particle equilibration processes are weak due to exponential suppression @xmath122 of the scattering near bottom of the band .
@xcite in this regime @xmath282 and the resistance of the wire , eq . , is given by @xmath283 .
it then follows from eq .
that to leading order in @xmath279 @xmath284 where @xmath285 has the meaning of dimensionless resistivity of the wire .
it is interesting to compare this result with eq .
obtained for a uniform wire . in the limit
@xmath52 using eqs .
and we extract from eq . the correction to resistance @xmath286 which coincides with eq . .
however , there is an important difference in the applicability of this result to uniform and disordered wires . in the case of uniform wires
@xmath286 applies only in the short wire limit @xmath52 , or equivalently at temperatures @xmath287 . for longer wires
@xmath241 saturates to the length - independent value @xmath288 , and thus always remains smaller than contact resistance , @xmath289 .
in contrast , in the case of disordered wires the result , eq . , does not rely on the assumption that @xmath52 .
the crossover temperature , eq .
, is controlled by disorder and does not depend on @xmath43 . thus , although the resistivity @xmath285 is small , for a sufficiently long wire the total resistance @xmath290 can be large , @xmath291 . at higher temperatures @xmath292 ( strong equilibration ) the resistance is given by the sum of two terms , @xmath293 the contribution of the momentum - nonconserving two - body collisions induced by disorder , the second term in eq . , dominates in the temperature regime @xmath294 , where @xmath295}\,.\ ] ] in this case the resistance of the wire is given by @xcite @xmath296 here we used eqs . and ,
set @xmath297 to the leading order in @xmath279 , and @xmath298 implies averaging along the wire . in refs . and the same result for resistivity
was derived assuming that electrons are fully equilibrated . here
we find that in fact applicability conditions for @xmath299 are more strict and eq
. dominates only in the temperature range @xmath300 . at higher temperatures
the resistance is governed by the first term in eq .
, @xmath301 where @xmath302 since @xmath159 . in eq .
we introduced thermal conductivity of a wire @xmath303 . indeed , according to eq . for long wires , @xmath54 , we have @xmath304 .
equation shows that resistance @xmath241 is determined by the magnitude of the disorder potential rather than its gradients .
similar to the case of two wires in series , considered in sec .
[ sec - transport - twowires ] , this feature can be traced back to the fact that in the inhomogeneous wire electrons never reach full equilibration no matter how long the wire is .
it is also interesting to note that resistance of the wire in the regime of strong equilibration [ eq .
] can be understood from purely hydrodynamic considerations .
@xcite enhanced equilibration due to three - particle collisions in the wire segment of length @xmath305 where the inhomogeneity potential is maximal . a larger value of @xmath306 near @xmath307 favors stronger electron backscattering in accordance with eq . .
in contrast , momentum - non - conserving two - particle collisions occur throughout the wire .
, width=302 ] we now relax the assumption of small variations in the amplitude of @xmath82 .
consider the case when generally smooth profile of the inhomogeneity potential has a well - defined maximum @xmath308 at some point @xmath309 inside the wire , see fig .
[ fig - barrier ] .
we assume that such strong fluctuation of @xmath82 happens in only one place along the wire .
this situation is likely to occur in practice due to the presence of charged impurities in the substrate and/or uneven screening of the nearby gates .
since @xmath306 is largest near @xmath309 and @xmath310 , see eq .
, it is natural to expect that three - particle equilibration processes are significantly enhanced in the part of the wire where @xmath82 reaches its maximum .
thus this region of the wire gives dominant contribution to resistance at lowest temperatures when @xmath311 .
assuming smoothness of @xmath312 at @xmath307 as compared to the sharp @xmath306 we can compute the resistance @xmath313by applying the saddle point approximation to eq . , @xmath314 here @xmath315 , @xmath316 and @xmath317 is the thermal length associated with the curvature of the potential @xmath82 near @xmath309 .
in contrast to the equilibration processes , dominated by scattering near @xmath309 , momentum - non - conserving two - particle collisions occur throughout the wire .
resistance of the wire is controlled by the latter at intermediate temperatures when @xmath318 .
for this regime @xmath319 which is analogous to eq . .
the only difference is that due to strong variations of @xmath82 spatial averaging in eq .
involves not only @xmath320 . at higher temperatures , when @xmath321 , resistance is again dominated by the scattering processes near the top of the inhomogeneity potential .
in this regime the first term in eq . , determined by the amplitude fluctuations of @xmath82 rather than its gradient , gives the leading contribution @xmath322 here we used @xmath323 found within saddle - point approximation from eqs . and to leading order in @xmath324 .
equation is analogous to eq . , with the thermal length @xmath305 effectively playing the role of the system size .
in this paper we studied the transport properties of weakly interacting one - dimensional electrons in the presence of inhomogeneities . in this system equilibration
is strongly restricted by the phase space available for electron scattering and conservation laws .
the resulting equilibration length @xmath45 is exponentially large at low temperatures and the partially equilibrated state is more likely to be realized than the fully equilibrated one .
furthermore , inhomogeneities present in the wire themselves resist equilibration of electrons due to momentum - non - conserving two - particle collisions . our main results are expressions for the resistance , peltier coefficient , thermal conductance and thermopower .
we find that the combined effect of interactions and inhomogeneities can dramatically increase the resistance of the wire .
for the long enough wire the induced correction could be much greater than the contact resistance of noninteracting electrons @xmath325 .
this is in contrast to the uniform case where interaction - induced correction to resistance saturates for @xmath54 and remains small as @xmath326 compared to the resistance @xmath325 of a non - interacting wire .
the combined effect of interactions and inhomogeneities is different for thermoelectric coefficients . on the one hand ,
when temperature increases , peltier coefficient and thermopower grow from exponentially small values , eq .
, to @xmath327 . on the other ,
this enhancement is not as dramatic as in the case of resistance .
indeed , the difference between the saturated values of peltier coefficient and thermopower in inhomogeneous wire as compared to that in the uniform wires is only in the appearance of the renormalized factor @xmath328 instead of the inverse chemical potential @xmath329 , see eq . .
conversely , the thermal conductance of the wire decreases due to the equilibration from its noninteracting value , eq . , to zero at @xmath54 .
the lack of complete electronic equilibration in the inhomogeneous quantum wires , which is another central observation of our work , warrants additional discussion .
the notion of equilibration appears naturally since initially right- and left - moving electrons entering the wire from the left and right lead , respectively , are at different equilibria with respect to each other due to the applied bias or temperature difference . in the case of weak interactions
three - particle collisions constitute the leading - order relaxation process . although the corresponding relaxation rate is slow ( or equivalently relaxation length is large ) due to the required scattering through the bottom of the band , complete equilibration between the right and left movers is nevertheless possible in a _ homogeneous _ wire .
once the length of the wire becomes large such that exponential suppression of the equilibration effects is compensated by a large system size , @xmath330 , the relaxation of the electron system becomes significant .
right and left movers eventually equilibrate to the single distribution , eq . .
when viewed in a reference frame moving together with electrons this distribution is simply the equilibrium fermi function .
thus , in the stationary frame this is distribution with a boost . in the presence of spatial _ inhomogeneities _ full equilibration is impeded .
this is most transparent when momentum - nonconserving two - body collisions are present , which unlike three - particle processes favor electron distribution function without boost , eq . .
as a result , the electron system is frustrated due to competition between two scattering processes and an intermediate distribution , eq . ,
is established .
interestingly , lack of full equilibration is a more general characteristic of inhomogeneous quantum wires , which is a consequence of conservation laws .
more precisely , in inhomogeneous wires it is generally not possible to reconcile full equilibration with conservation of energy . technically speaking
, this observation comes from the fact that in the fully equilibrated state all currents are proportional to the drift velocity . in the linear response regime one
therefore has only a single parameter @xmath74 to simultaneously satisfy uniformity of particle and heat currents along the wire , imposed by conservation laws . in inhomogeneous wires
this is generally not possible since for @xmath264 the ratio @xmath331 is not constant along the wire .
thus the electron liquid must remain in the state of partial equilibration with @xmath219 .
we have illustrated this point explicitly by considering the simplest example of inhomogeneity , namely , a junction of two uniform wires with mismatched densities . in the general case of a wire with long - range disorder
the consequence of the partially equilibrated state is that interaction - induced correction to resistance of the wire is determined by the amplitude of the variations in inhomogeneity potential , rather than its gradients .
this correction may be large compared to the resistance @xmath325 of non - interacting wires .
we are grateful to b. l. altshuler , a. v. andreev , a. p. dmitriev , y. m. galperin , i. v. gornyi , d. g. polyakov and b. shklovskii for helpful discussions .
this work at anl was supported by the u.s .
department of energy , office of science , under contract no .
de - ac02 - 06ch11357 .
[ sec - appendix ]
for the inhomogeneous wire we describe the electron - electron interaction responsible for two - body scattering by its general translationally non - invariant form , @xmath332 as a function of its first argument @xmath217 is assumed to be coulombic in nature and thus short - ranged , with variations on the scale of a certain screening length due to the nearby gates .
the inhomogeneity is captured by the second argument with a corresponding variations in @xmath217 on the length scale @xmath17 , large compared to both , the fermi wavelength @xmath79 and the range of screening .
our starting point for the energy - transfer rate from the right movers for the segment of the wire @xmath333 is the following golden - rule expression , @xmath334\,,\end{aligned}\ ] ] where the matrix element @xmath335 into the final state @xmath336 .
these matrix elements will be calculated on the basis of semiclassical wave functions , @xmath337}\right]\,,\ ] ] which are eigenstates of the free hamiltonian @xmath338\psi_{\epsilon}(x)=\epsilon\psi_{\epsilon}(x)$ ] normalized according to @xmath339 and @xmath340 .
the subscripts @xmath341 refer to the right / left branches and we also ignored the backscattering wave , since it only leads to an exponentially small contribution for @xmath342 . focusing on the states close to the fermi energy , we can simplify the expression for the eigenstates , eq .
, of the free hamiltonian into @xmath343\,,\ ] ] where @xmath344 is obtained from eq . by setting @xmath345 .
this allows us to find , for example , the matrix element @xmath346 to first order in the interaction @xmath347 ^ 2 } \nonumber\\ & & \hskip-1.25cm\times\int^{\delta x}_{-\delta x}dy\ , \mathcal{v}(y , x)\left(1-e^{-2ik_{f}(x)y}\right)\,.\end{aligned}\ ] ] here we introduced center of mass @xmath348 and relative @xmath349 coordinates , and @xmath350 . using eq . and expanding the occupation factors @xmath351 in eq . to linear order in @xmath148 , and splitting the energy - conserving delta function into two as @xmath352
, we can complete @xmath353 and @xmath354 integrations and find @xmath355 where @xmath356 is now the equilibrium fermi function .
the corresponding matrix element in these notations reads @xmath357 ^ 2}\nonumber \\ \times\exp\left(-2i\omega\int^{x}_{0}\frac{dx'}{\hbar v_{f}(x')}\right)\,.\end{aligned}\ ] ] the shortened forms @xmath358 and @xmath359 correspond to the zero momentum and @xmath216 fourier components of the potential @xmath360 with respect to its first variable @xmath361 defined as @xmath362 and @xmath363 . at this stage @xmath128 and @xmath364
integrations can be completed by noticing that @xmath365=\pm\omega\,.\end{aligned}\ ] ] as a result , we obtain following expression for the energy - transfer rate : @xmath366\left[\frac{\mathcal{v}_{0}(x_2)-\mathcal{v}_{2k_f}(x_2)}{\pi\hbar v_{f}(x_2)}\right ] \nonumber\\ & & \hskip-1cm\int^{+\infty}_{-\infty}d\omega\left(\frac{\omega^2/t}{\sinh\frac{\omega}{2t}}\right)^2 \frac{\exp\left(-2i\omega\int^{x_2}_{x_1}\frac{dx'}{\hbar v_f(x')}\right)}{\hbar v_{f}(x_1)\hbar v_{f}(x_2 ) } \,.\end{aligned}\ ] ] being interested in the temperature range @xmath367 , where the exponential is rapidly oscillating , we write @xmath368 integrate by parts over @xmath369 and @xmath370 and complete the remaining energy integral , which gives @xmath371 . due to the delta function , one spatial integration is thus removed and we find as the final result , eq .
, where all three scattering channels were included .
the resistance of a junction between two uniform wires with different densities can be found simply by combining conservation laws for the currents with the microscopic equations for @xmath102 and @xmath78 .
our starting point is eq . which is naturally generalized to the case of two wires connected in series @xmath372 where @xmath373 correspond to the change in the number of right - movers within each segment of the wire .
similarly , generalization of eq
. for the heat balance reads @xmath374 equation defines the resistance , eq .
, where interactions - induced correction is given by @xmath375 we use now eq . in the limit of uniform wires to express rates @xmath373 as @xmath376 by introducing the new quantity @xmath377 , which has meaning of the degree of equilibration .
indeed , @xmath378 corresponds to the limit of no equilibration , such that the current is determined solely by @xmath221 , while @xmath379 corresponds to full equilibration where @xmath47 .
this notation makes it possible to express @xmath380 in eq . in terms of @xmath381 as @xmath382 , which was used in eq . .
conservation of the heat current , eq .
, @xmath383 imposes a constraint @xmath384 , which must be constant . with the help of eq . the expression for @xmath241 in eq .
can be rewritten as @xmath385 where the unknown quantity @xmath386 is yet to be determined . the way to find it is from the energy conservation .
recall , that @xmath78 and @xmath102 are related to each other by eq . since they are caused by the same scattering mechanism . for the uniform wire eq
. reads @xmath387 . as a result , for eq . in the case of no temperature bias
@xmath388 we have @xmath389 inserting here eqs .
and one finds @xmath390 which allows to find @xmath386 explicitly , @xmath391 finally , using this @xmath386 in eq . and
reexpressing @xmath392 through the equilibration length as @xmath393 one recovers eq . .
we thus conclude that eq .
is just a consequence of the conservation laws .
d. j. reilly , g. r. facer , a. s. dzurak , b. e. kane , r. g. clark , p. j. stiles , r. g. clark , a. r. hamilton , j. l. obrien , n. e. lumpkin , l. n. pfeiffer and k. w. west , phys .
b * 63 * , 121311(r ) ( 2001 ) . in the limit , where the equilibration due to triple collisions is weaker than that due to pair collisions in the presence of forward - scattering disorder , the resistivity was also studied by d. n. aristov , a. p. dmitriev , i. v. gornyi , and d. g. polyakov , unpublished .
|
we study transport properties of weakly interacting one - dimensional electron systems including on an equal footing thermal equilibration due to three - particle collisions and the effects of large - scale inhomogeneities .
we show that equilibration in an inhomogeneous quantum wire is characterized by the competition of interaction processes which reduce the electrons total momentum and such which change the number of right- and left - moving electrons .
we find that the combined effect of interactions and inhomogeneities can dramatically increase the resistance of the wire .
in addition , we find that the interactions strongly affect the thermoelectric properties of inhomogeneous wires and calculate their thermal conductance , thermopower , and peltier coefficient .
|
various types of detectors require algorithms that handle number of datapoints in a plane and reconstruct the parameters of a circle , @xcite,@xcite,@xcite .
these algorithms need to be robust against noise hits and analyse big amount of data relatively fast .
finally , the performance of the algorithm needs as independent as possible with the resolution of the detector . in a recent paper @xcite we examined the use of the legendre transform for the determination of a circle s characteristics . in this paper
we revisit this problem and describe in detail an alternative method , based on gaussian sums , proposed for the reconstruction of the circle s center @xmath0 and radius @xmath1 from a given set of datapoints .
following the notation and train of though of @xcite , using @xmath2 datapoints ( @xmath3 , @xmath4 and @xmath5 ) , we can have a first estimation of the center @xmath6 and the radius , @xmath7 , of the circle ( see figure [ fig : circle_circle ] ) . and radius @xmath7 . ]
these estimates are given by : @xmath8 where @xmath9 and @xmath10 the slopes of the lines connecting @xmath11 and @xmath12 .
having @xmath13 given datapoints / measurements , @xmath14 different centers and radii can be estimated . using the formulas described above the global sums
are constructed @xmath15 where @xmath16 , @xmath17 and @xmath18 are the circle s parameters for the @xmath19 triplet of datapoints that defines the circle .
the standard deviations of the gaussians are set of a fraction of the geometrical scale of the problem . in the studies below @xmath20
have been used .
the reason of using gaussian functions in the definition of the global sums ( see above ) instead of using the basic @xmath21 test functions , is based on the fact that noise type datapoints will give values of the parameters far away from the right answer . these estimated parameters ( @xmath16 , @xmath17 and @xmath18 ) will contribute very little to the global sums . on the contrary
, the right values of the estimated parameters will give the maximum contribution .
for each event the functions described above ( see figure [ fig : gaussians_sum_example ] ) are constructed and the value of the bin with the maximum number of entries is chosen .
the value of these bins are the estimate of the @xmath22 and @xmath7 of the circle .
the extraction of the circle parameters using the first method ( see section [ sec : model_description_first ] ) is straightforward : the bin with the maximum number of entries per event in each @xmath23 , @xmath24 and @xmath25 gives the estimate of the center s coordinates and the radius of the circle .
an example event can be seen in figure [ fig : gaussians_sum_example ] . in figure
[ fig : one_circle ] the circle used to generate the datapoints is shown .
the @xmath25 , @xmath23 and @xmath24 distributions for the circle parameters can be seen in figures [ fig : one_circle_r_gauss ] , [ fig : one_circle_x_gauss ] and [ fig : one_circle_y_gauss ] respectively . @xmath23 and @xmath24 for @xmath26 and @xmath27 peak at @xmath28 where the center of the circle is and @xmath25 peaks at @xmath29 indicating the actual radius of the circle .
the algorithm is tested using simulated experiments , creating datapoints ( @xmath30 ) that belong to a circle .
the robustness of the algorithms was tested by inserting gaussian uncertainties in both the @xmath31 and @xmath32 coordinates of the datapoints / measurements and random noise hits / outliers .
a special scenario was considered , where one quadrant of the detector was deactivated ( results are summarized in table [ tab : results_no_quad ] ) .
the center of the circle was chosen to be at @xmath33 with a radius of @xmath34 ( arbitrary units ) . using this circle ,
@xmath13 datapoints ( @xmath35 and @xmath36 ) are randomly created .
we smear both the @xmath37 and @xmath38 positions of the datapoints , independently on @xmath31 and @xmath32 , by @xmath39 , @xmath40 or @xmath41 of the circle s radius @xmath1 .
apart from the uncertainty introduced in the position of the datapoints , noise hits in certain percentage of the datapoints that originate from the circle are included .
the noise hits are randomly generated in the ranges @xmath42 and @xmath43 .
an event generated by this toy mc can be seen in figure [ fig : one_circle ] .
the red and green datapoints ( @xmath44 ) are generated from the circle and the black datapoints are the noise hits ( @xmath45 noise level is used in this event ) .
the positions of all the generated datapoints @xmath46 are smeared by @xmath40 of its radius @xmath1 .
the results of the methods described in section [ sec : model_description ] are summarized in table [ tab : results ] .
the algorithm proposed for the extraction of the circle s parameters work well under all test scenarios .
.results of the algorithms described in section [ sec : sub_extracion ] .
we use @xmath47 datapoints , generated in three quadrants of the circle , three different percentages of position uncertainty ( @xmath39 , @xmath40 and @xmath41 of the radius @xmath1 ) and five additional noise hits .
the original values of the circle s parameters can be found in the first row . [
cols="^,^,^,^",options="header " , ] [ tab : results_no_quad ] + noise [ % ] & & & + uncertainty [ % ] & 5 & 10 & 15 & 5 & 10 & 15 & 5 & 10 & 15 + r & @xmath48 & @xmath49 & @xmath50 & @xmath49 & @xmath51 & @xmath50 & @xmath49 & @xmath51 & @xmath52 + x & @xmath53 & @xmath54 & @xmath54 & @xmath53 & @xmath54 & @xmath54 & @xmath53 & @xmath54 & @xmath55 + y & @xmath56 & @xmath57 & @xmath58 & @xmath56 & @xmath57 & @xmath59 & @xmath56 & @xmath59 & @xmath59 + + noise [ % ] & & & + uncertainty [ % ] & 5 & 10 & 15 & 5 & 10 & 15 & 5 & 10 & 15 + r & @xmath49 & @xmath49 & @xmath60 & @xmath48 & @xmath51 & @xmath61 & @xmath49 & @xmath62 & @xmath52 + x & @xmath53 & @xmath53 & @xmath53 & @xmath53 & @xmath63 & @xmath64 & @xmath53 & @xmath54 & @xmath64 + y & @xmath56 & @xmath65 & @xmath57 & @xmath56 & @xmath56 & @xmath57 & @xmath56 & @xmath57 & @xmath57 + + noise [ % ] & & & + uncertainty [ % ] & 5 & 10 & 15 & 5 & 10 & 15 & 5 & 10 & 15 + r & @xmath49 & @xmath49 & @xmath49 & @xmath66 & @xmath49 & @xmath51 & @xmath49 & @xmath51 & @xmath51 + x & @xmath53 & @xmath53 & @xmath54 & @xmath53 & @xmath53 & @xmath54 & @xmath53 & @xmath67 & @xmath64 + y & @xmath56 & @xmath56 & @xmath59 & @xmath56 & @xmath56 & @xmath57 & @xmath56 & @xmath56 & @xmath57 + + noise [ % ] & & & + uncertainty [ % ] & 5 & 10 & 15 & 5 & 10 & 15 & 5 & 10 & 15 + r & @xmath60 & @xmath49 & @xmath51 & @xmath60 & @xmath49 & @xmath62 & @xmath68 & @xmath51 & @xmath62 + x & @xmath53 & @xmath53 & @xmath69 & @xmath53 & @xmath53 & @xmath53 & @xmath53 & @xmath53 & @xmath54 + y & @xmath57 & @xmath56 & @xmath57 & @xmath65 & @xmath56 & @xmath57 & @xmath56 & @xmath56 & @xmath57 + [ tab : results ]
we examined the use of gaussian sums for the reconstruction of a circle s center and radius , using its datapoints .
the algorithm proposed proved to work well under all test scenarios ( noise hits , inactive part of the detector , hit position uncertainties ) and provide fast and accurate results .
the research leading to these results has received funding from the fp7-infrastructures-2010 , aida : advanced european infrastructures for detectors at accelerators , grant agreement no : 262025 and the european union ( european social fund esf ) and the greek national funds through the operational program `` education and lifelong learning '' of the national strategic reference framework ( nsrf ) 2007 - 1013 . aristeia-1893-atlas micromegas .
the support is gratefully acknowledged .
altherr and j.seixas ] , cherenkov ring recognition using a non - adaptable network , nuclear instruments and methods in physics research a317 ( 1992 ) 335 - 338 http://download.springer.com/static/pdf/419/art%253a10.1007%252fs10582-000-0075-0.pdf?auth66=1394193431_257a1dc89574bf1dba670e5980ddb1ba&ext=.pdf[p .
volf et al .
] , on mathematical approach to circle fitting problem in the analysis of rich detector data , czech .
50/s1 ( 2000 ) http://ac.els-cdn.com/037704279500274x/1-s2.0-037704279500274x-main.pdf?_tid=c858731c-a45c-11e3-ba31-00000aab0f6c&acdnat=1394020595_296f639b5e45e1b23924e4ad480fd712[h .
babovsky ] , recognition of ring pairs in the data analysis of cherenkov detectors , journal of computational and applied mathematics 72 ( 1996 ) 227 - 234 http://ac.els-cdn.com/s0168900214000977/1-s2.0-s0168900214000977-main.pdf?_tid=211a96d2-a44f-11e3-9543-00000aab0f6b&acdnat=1394014731_ba2c7350bb64881524afcd63b7e9b40b[t .
alexopoulos et al . ] , identification of circles from datapoints using the legendre transform , nuclear instruments and methods in physics research a745 ( 2014 ) 16 - 23
|
we present a pattern recognition method which use datapoints on a plane and estimates the parameters of a circle .
mc data are generated in order to test the method s efficiency over noise hits , uncertainty in the hits positions and number of datapoints .
the scenario were the hits from a quadrant of the circle are missing is also considered .
the method proposed is proven to be robust , accurate and very efficient .
|
the ionization of atoms in an external electric field is , as well - known , one of the oldest subjects in quantum physics . as a simple example of the ionization , hydrogen - like atoms in a uniform electrostatic field
have been extensively considered ( see , e.g. , @xcite ) . here
the background potential caused by the field decreases without limit in one direction , and the electron initially in the bound state will eventually tunnel through the `` barrier '' created by the field , so leading to ionization of the atom . the tunneling rate for an _ ensemble _ of many independent electrons has been calculated based on the exponential decay law following from the statistical assumption that the tunneling rate is proportional to the number of available atoms . on the other hand , big experimental advances in the field of nano - scale physics have highly enlarged a need for a detailed understanding of the ( time - dependent ) tunneling process of _ individual _ electrons subjected to an external field .
in fact , the nano - scale devices have been designed and fabricated , examples of which are molecular switches @xcite and resonant tunnel junctions @xcite , etc . however , in exploring the time evolution of the electron tunneling process leading to ionization , we have a considerable mathematical difficulty that there are no exactly solvable models for a transition from a bound state to the continuum .
further , even obtaining the numerical solution , with high accuracy , to this problem can not be considered an easy task either , especially in the strong - field limit where a highly oscillatory behavior is found in the time evolution of the bound - continuum transition .
the system under investigation in this paper is a particle subjected to an attractive one - dimensional delta - function potential , @xmath6 .
we intend to study analytically the time evolution of the particle when a uniform electrostatic field @xmath0 is applied .
the delta - function potential well ( not in an external field ) has a single bound state .
this aspect may rather simplify the analytical study of the ionization .
however , no exact solution @xmath2 to the time - dependent schrdinger equation of this problem has been found in closed form ( in terms of the actual calculability to any sufficient degree of precision ) , even for such a simple form of external field .
in fact , this model of the field - induced time - dependent ionization has already been studied by some people .
it was first discussed by geltman @xcite .
later on , several different approaches to obtaining the time - dependent wavefunction @xmath2 have been carried out @xcite .
one of the approaches mainly adopted so far is to turn the relevant time - dependent schrdinger equation into the lippmann - schwinger integral equation [ cf .
( [ eq : lippmann_schwinger_eq1 ] ) and ( [ eq : lippmann_schwinger_eq0 ] ) ] .
this integral equation is , however , highly non - trivial to solve analytically and so has been focused mostly upon its numerical solvability . in @xcite by elberfeld and kleber , interestingly enough , the time - dependent ionization probability has been investigated in the strong - field limit based on the numerical analysis of the integral equation .
they also demonstrated numerically that the ionization probability obtained from the simple exponential decay law may be considered a fairly good approximation on the average to the exact result in the strong - field limit although this approximation , by construction , can not account for the short - time ripples , observed in the exact one , which result from the ionization - recombination process in the bound - continuum transition @xcite .
the goal of this paper lies in a systematic derivation of the propagator @xmath1 of the system , directly leading to @xmath7 . as our central finding , its essential constituent part @xmath3 , accounting for the ionization - recombination ,
is exactly expressed in terms of the multiple hypergeometric functions @xmath4 , and then in an appropriate approximation scheme in terms of the generalized hypergeometric functions @xmath5 being much more transparent to physically interpret and much more accessible in their numerical evaluation than the functions @xmath4 . the general layout is the following . in sec .
[ sec : review ] we briefly review the well - known results of both the delta - function potential problem without an external field and the problem of a particle subjected to a uniform electrostatic field but not bound by the ultrathin potential well . in doing so , we also sophisticate the old results . in sec .
[ sec : propagator ] , we derive an explicit expression of the propagator in terms of the multiple hypergeometric functions and then discuss its mathematical complexity . sec .
[ sec : approximation ] we obtain the propagator in approximation , expressed in terms of the generalized hypergeometric functions .
finally , we give the conclusion of this paper in sec . [
sec : conclusion ] .
the system under consideration is described by the hamiltonian @xmath8 where an external field @xmath9 , and @xmath10 is a strength of the @xmath11-potential well . for the field - free case ( @xmath12 ) ,
it is well - known that the hamiltonian @xmath13 has a single bound state @xcite @xmath14 with its eigen energy @xmath15 where @xmath16 .
all eigenstates and eigenvalues of @xmath17 as well as the completeness of the eigenstates have been discussed in detail in @xcite . and
for the potential - well - free case ( @xmath18 ) , the eigenfunction of the hamiltonian @xmath19 with ( continuous ) energy @xmath20 is given by @xcite @xmath21 in terms of the airy function @xmath22 .
the method we employ here to obtain the propagator of the system @xmath23 is to apply the relevant lippmann - schwinger integral equation @xcite @xmath24 where the propagator @xmath25 . as well - known , the propagator has the physical meaning of a ( complex - valued ) transition - probability amplitude to get from the point @xmath26 to the point @xmath27 @xcite . here
@xmath28 represents the propagator relevant to the partial hamiltonian @xmath29 only , explicitly given by @xcite @xmath30 in which the propagator @xmath31 for a free particle subjected to @xmath32 only . also , @xmath33 is a field - induced classical impulse , leading to the corresponding field - induced translation @xmath34 and the field - induced action @xmath35 .
( [ eq : lippmann_schwinger_eq1 ] ) immediately gives rise to @xmath36 where the homogeneous solution @xmath37 obviously represents free motion subjected to an external field @xmath0 only , while the second term on the right - hand side , by construction , gives the influence of the residual zero - range potential .
as an example , for the initial bound state @xmath38 , eq .
( [ eq : homogeneous_solution1 ] ) reduces to a closed expression @xcite @xmath39 in terms of the moshinsky function @xcite @xmath40 where @xmath41 is the complementary error function @xcite .
it is also instructive to point out that we can employee , as an alternative to the interaction hamiltonian @xmath42 in ( [ eq : total_hamiltonian1 ] ) given in the scalar - potential gauge , its counterpart @xmath43 in the vector - potential gauge , and the system of interest is accordingly given by @xmath44 where the vector potential @xmath45 corresponds to the field - induced impulse @xmath46 .
then it can easily be verified that @xmath47 substituting ( [ eq : vector - potential - gauge - terms11 ] ) and ( [ eq : vector - potential - gauge - terms12 ] ) into ( [ eq : lippmann_schwinger_eq1 ] ) and ( [ eq : lippmann_schwinger_eq0 ] ) respectively , we can straightforwardly obtain the equivalent results in the vector - potential gauge to what below follows in the scalar - potential gauge ( cf . for a detailed discussion of @xmath42 versus @xmath43 gauge problem ,
see , e.g. , @xcite ) . besides
, our formalism can apply to the system of a delta - potential barrier under an electrostatic field @xmath48 as well , simply by replacing @xmath49 by @xmath50 in eq .
( [ eq : lippmann_schwinger_eq1 ] ) or ( [ eq : lippmann_schwinger_eq0 ] ) and then going forward straightforwardly .
in fact , the eigenstates and eigenvalues of the hamiltonian @xmath51 are identical to those of @xmath52 , except for the fact that @xmath53 does not have the bound state @xmath54 as its eigenstate . to develop the forthcoming discussions in a simplified fashion , let us rescale coordinate and time as the dimensionless quantities , @xmath55 and @xmath56 , respectively @xcite , which is basically equivalent to setting @xmath57 .
then the schrdinger equation for the hamiltonian @xmath58 easily reduces to @xmath59 where the relative field strength @xmath60 , and eq .
( [ eq : lippmann_schwinger_eq1 ] ) accordingly appears as @xmath61 where the rescaled propagator @xmath62 which is identical to @xmath63 .
for the sake of convenience we replace the notation @xmath64 by @xmath65 for what follows . in comparison , we also review briefly the exact results of the field - free case ( @xmath66 ) .
we first set @xmath67 in ( [ eq : lippmann_schwinger_eq2 - 0 ] ) , which then enables @xmath68 on the left - hand side to immediately substitute for @xmath69 on the right - hand side .
making iterations of the substitution , we can finally arrive at the expression @xcite @xmath70 where @xmath71 and @xmath72 . here
the index @xmath73 represents a number of the substitutions and so that of interactions with the @xmath11-potential well , and @xmath74 the repeated erfc integral [ cf .
@xmath75 . using the identity @xcite @xmath76 we can easily obtain @xmath77 which subsequently leads to the closed expression @xcite @xmath78
let us come back to the field - induced dynamics ( @xmath79 ) in order to explore an explicit expression of @xmath80 as the counterpart to the result in ( [ eq : field - free8 ] ) .
for doing so , it may be useful to apply the laplace transform @xmath81 to eq .
( [ eq : lippmann_schwinger_eq2 - 0 ] ) with respect to time ; we do this job first for @xmath67 with the help of the convolution theorem , which will finally give rise to @xcite @xmath82 where the laplace transform @xmath83 . from eq .
( [ eq : kernel2 ] ) it can easily be shown as well that @xmath84 . to obtain an explicit form of @xmath85
, we use the integral representation @xcite @xmath86 where @xmath87 and @xmath88 are real - valued , and the airy function @xmath89 .
this allows us to have the green function in the frequency domain @xcite @xmath90 where @xmath91 and @xmath92 .
therefore , a closed expression of the green function @xmath93 , identical to @xmath94 in eq .
( [ eq : general - propagator6 ] ) with @xmath95 , immediately appears in terms of the airy functions in ( [ eq : general - propagator9 ] ) .
however , it is highly non - trivial to directly carry out the inverse fourier transform of the closed expression of @xmath96 in the frequency domain , even with the stationary phase approximation , in order to obtain an explicit expression of @xmath80 in the time domain @xcite . for a later purpose ,
we introduce here the time - dependent ionization probability for the initial bound state @xmath54 , which is defined as @xmath97 here the bound state amplitude @xmath98 where , from eq .
( [ eq : lippmann_schwinger_eq0 ] ) , @xmath99 substituting ( [ eq : homog_sol1 ] ) into ( [ q : second - term - of - ionization1 ] ) , we can easily obtain a closed form @xcite @xmath100 which reduces to unity at @xmath101 , as required
. then it has been shown @xcite that in the weak - field limit @xmath102 , the exponential decay law @xmath103 is a good approximation on the average .
this approximation can also be simulated by the ansatz in `` smooth '' form @xmath104 where the complex - valued energy @xmath105 with @xmath106 and the decay rate @xmath107 @xcite . here , @xmath108 is the level shift .
the semiclassical value @xmath109 is a good approximation of @xmath110 for @xmath111 , and @xmath112 is in excellent agreement to @xmath108 up to @xmath113 .
however , this exponential decay approximation , by construction , can not account for the ripples observed in the exact time evolution of the ionization probability resulting from the ionization - recombination process in the bound - continuum transition , explicitly demonstrated in @xcite from the numerical treatment of the lippmann - schwinger equation .
to derive an explicit expression of the propagator , we make the same iterations to eq .
( [ eq : lippmann_schwinger_eq2 - 0 ] ) as those applied above to the field - free case leading to ( [ eq : field - free6 ] ) , which then reveals that @xmath114 to simplify the notation , let @xmath115 and then @xmath116 from now on [ cf .
( [ eq : kernel2 ] ) ] . for a later purpose , we rewrite eq .
( [ eq : propagator - expand1 ] ) as @xmath117 where the partial integrand , being a function of time only , @xmath118 eq .
( [ eq : propagator - expand - field1 - 0 ] ) can easily be verified with the help of the laplace transform in ( [ eq : general - propagator6 ] ) , but without considering eq .
( [ eq : general - propagator9 ] ) ; the laplace transform is straightforwardly expanded as @xmath119 and then we simply apply the inverse laplace transform to this expression term by term , which readily recovers the result in ( [ eq : propagator - expand - field1 - 0 ] ) . from this
, it also follows that @xmath120 as seen , the finding of an explicit expression of the quantity @xmath121 is a critical step to deriving the wavefunction @xmath122 for any initial state @xmath123 [ cf .
( [ eq : lippmann_schwinger_eq0 ] ) ] .
comments deserve here .
first , the substitution of @xmath124 into eq .
( [ eq : propagator - expand - field1 - 0 ] ) and then the comparison with ( [ eq : propagator - field1 - 1 - 0 ] ) gives rise to a noteworthy relation @xmath125 which will be useful later .
secondly , the quantity @xmath126 in ( [ eq : laplace_transform1 ] ) designates the green function for a closed path with the initial and final point being @xmath67 , and the summation index @xmath73 represents how many times the closed path interacts with the potential well located at @xmath67 .
then the quantity @xmath121 , exactly corresponding to this summation in the time domain , is accordingly responsible for the ionization - recombination resulting from the scattering from the zero - range potential at @xmath67 , particularly for @xmath127 , which was qualitatively explained in @xcite as follows : this initial bound state , explicitly given by @xmath128 in the momentum representation , has a symmetric momentum distribution around @xmath129 for the particles ( in an ensemble represented by the bound state ) . by applying the field ,
this symmetry breaks down in such a way that the motion of the particles in one direction is accelerated and so they will easily leave the potential well , simply toward the ( continuous ) unbound states . on the other hand , the motion in the other direction
gets slowed down until the particles stop , and then they reverse their direction of motion so that the particles again approach the potential well at @xmath130 where they cause a transient maximum in the bound state probability .
this process is repeated until all particles will completely leave the potential well .
below we will systematically derive the quantity @xmath121 , and so @xmath131 , in closed form .
let first @xmath132 and @xmath133 in ( [ eq : propagator - field1 - 1 - 0 ] ) .
we consider the next term @xmath134 , where the integral @xmath135 [ cf .
( [ eq : kernel2 ] ) ] .
let @xmath136 . applying to eq .
( [ eq : propagator3 ] ) the taylor series of the exponential function and then the identity of beta function @xcite @xmath137 where @xmath138 , it easily appears that @xmath139 with the help of the gamma function identities , @xmath140 and @xmath141 and @xmath142 @xcite , eq . ( [ eq : propagator20 - 0 ] ) reduces to a closed expression @xmath143 3\hspace*{-.1cm}:&\hspace*{-.2cm}-\hspace*{-.01cm};&\hspace*{-.2cm}- \end{array}\right.\left(\begin{array}{cccccc } -&:&\left[\frac{1}{6 } , \frac{1}{2 } , \frac{5}{6}\right]&;&\left[\frac{1}{6 } , \frac{1}{2 } , \frac{5}{6}\right]&;\\ \left[\frac{1}{3 } , \frac{2}{3 } , 1\right]&:&-&;&- & ; \end{array}\,a , a \right)\ ] ] with @xmath144 in terms of the generalized multiple hypergeometric function @xcite @xmath145 q_0\hspace*{-.13cm}:&\hspace*{-.2cm}q_1\hspace*{-.05cm};&\hspace*{-.2cm}\cdots\hspace*{-.05cm};&\hspace*{-.2cm}q_n \end{array}\right.\left(\begin{array}{cccccccc } { \bf a}_0&:&{\bf a}_1&;&\cdots&;&{\bf a}_n&;\\ { \bf b}_0&:&{\bf b}_1&;&\cdots&;&{\bf b}_n&;\\ \end{array}\,z_1 , \cdots , z_n \right ) : = \sum_{k_1,\cdots , k_n = 0}^{\infty } \frac{({\bf a}_0)_{k_1+\cdots+k_n}}{({\bf b}_0)_{k_1+\cdots+k_n}}\ , \prod_{j=1}^n \frac{({\bf a}_j)_{k_j}}{({\bf b}_j)_{k_j } } \frac{z_j^{k_j}}{k_j!}\ ] ] where @xmath146 and @xmath147 with @xmath148 are vectors with dimensions @xmath149 and @xmath150 , respectively . and @xmath151 and @xmath152 , where the pochhammer symbol @xmath153 .
the multiple series in ( [ eq : multiple - hypergeometric0 - 1 ] ) absolutely converges if @xmath154 for all @xmath155 .
in fact , eq . ( [ eq : propagator20 - 1 ] ) fulfills this condition as @xmath156 for @xmath157 .
subsequently we consider the next term @xmath158 , where the integral @xmath159 similarly to @xmath160 , we can straightforwardly obtain @xmath161 let @xmath162 and @xmath136 .
we then use the identity @xcite @xmath163 where @xmath164 .
( [ eq : identity_gamma1 ] ) is easily shown to be identical to @xmath165 , where @xmath166 and @xmath167 .
consequently it appears that @xmath168 with the aid of the technique already applied to ( [ eq : propagator20 - 0 ] ) , this reduces to a closed expression @xmath169 3\hspace*{-.1cm}:&\hspace*{-.2cm}-\hspace*{-.01cm};&\hspace*{-.2cm}-\hspace*{-.01cm};&\hspace*{-.2cm}- \end{array}\right.\left(\begin{array}{cccccccc } -&:&\left[\frac{1}{6 } , \frac{1}{2 } , \frac{5}{6}\right]&;&\left[\frac{1}{6 } , \frac{1}{2 } , \frac{5}{6}\right]&;&\left[\frac{1}{6 } , \frac{1}{2 } , \frac{5}{6}\right]&;\\ \left[\frac{1}{2 } , \frac{5}{6 } , \frac{7}{6}\right]&:&-&;&-&;&- & ; \end{array}\,a , a , a \right)\ ] ] with @xmath144 . along the same line
, we can subsequently obtain closed expressions of @xmath170 and @xmath171 , respectively , where @xmath172 and @xmath173 for eq .
( [ eq : propagator - field1 - 1 - 0 ] ) .
now let us generalize the above scenario to the quantity @xmath174 , where the integral @xmath175 to do so , we apply the same technique with the aid of the identity @xcite @xmath176 where @xmath177 for @xmath178 .
this allows us to finally arrive at the expression @xmath179 which subsequently reduces to the closed expression @xmath180 3\hspace*{-.1cm}:&\hspace*{-.2cm}-\hspace*{-.01cm};&\hspace*{-.2cm}\cdots\hspace*{-.05cm};&\hspace*{-.2cm}- \end{array}\right.\left(\begin{array}{cccccccc } -&:&\left[\frac{1}{6 } , \frac{1}{2 } , \frac{5}{6}\right]&;&\cdots&;&\left[\frac{1}{6 } , \frac{1}{2 } , \frac{5}{6}\right]&;\\ \left[\frac{n}{6 } , \frac{n+2}{6 } , \frac{n+4}{6}\right]&:&-_{\scriptscriptstyle{(1)}}&;&\cdots&;&-_{\scriptscriptstyle{(n ) } } & ; \end{array}\,\underbrace{a , \cdots , a}_{\scriptscriptstyle{n } } \right)\,.\n\end{aligned}\ ] ] as a result , we find the exact expression @xmath181 in terms of the ( well - defined ) multiple hypergeometric functions . from ( [ eq
: lambda - propagator-0 - 0 ] ) , it easily follows as well that @xmath182 . by substituting eqs .
( [ eq : kernel2 ] ) and ( [ eq : propagator - field1 - 1 - 0 - 0 ] ) into ( [ eq : propagator - expand - field1 - 0 ] ) , we can now obtain an explicit expression of the propagator @xmath80 in terms of the single double integral @xmath183 ; this remaining double integral can not straightforwardly be evaluated indeed in terms of the multiple hypergeometric functions in ( [ eq : multiple - hypergeometric0 - 1 ] ) .
however , it is still desirable to express eq .
( [ eq : propagator - field1 - 1 - 0 - 0 ] ) ( in good approximation ) in terms of the generalized hypergeometric functions @xmath184 ( with a single argument @xmath185 ) @xcite being much more transparent to physically interpret and much more accessible in their numerical evaluation than the multiple hypergeometric functions [ cf .
( [ eq : propagator - field1 - 1 ] ) and ( [ eq : propagator - field2 - 1 ] ) ] .
in order to discuss the mathematical complexity of eq .
( [ eq : propagator - field1 - 1 - 0 - 0 ] ) , it is also instructive to formally express the quantity @xmath121 in terms of a single integral over the @xmath73-sphere solid angle @xmath186 ( e.g. , a @xmath187-sphere corresponding to a circle ) , in which the infinitesimal element @xmath188 , and the integral over the entire region is explicitly given by @xmath189 @xcite . to this end , we first consider the integral @xmath160 in ( [ eq : propagator3 ] ) .
let @xmath190 .
this immediately allows us to have @xmath191 which reduces to a closed form @xmath192 in terms of the bessel function @xmath193 [ cf .
( [ eq : propagator20 - 1 ] ) ] .
here we used @xmath194 and then @xmath195 @xcite .
the integral @xmath196 is next under consideration .
we substitute eqs .
( [ eq : kernel2 ] ) and ( [ eq : propagator5 ] ) into ( [ eq : propagator6 ] ) and then carry out the same change of integral variable as that used for ( [ eq : propagator4 ] ) . then it appears that @xmath197 [ cf .
( [ eq : propagator17 ] ) for exact evaluation of this integral ] .
we next pay an explicit attention to @xmath198 [ cf .
( [ eq : propagator23 ] ) with @xmath199 . using @xmath200 and then introducing @xmath201 , the first two - integral part of ( [ eq : propagator-7 - 0 ] ) is easily transformed into @xmath202 .
this , with eqs .
( [ eq : propagator3 ] ) and ( [ eq : propagator5 ] ) , then allows us to have @xmath203 applying again the very technique already used for @xmath160 and @xmath196 , we can finally obtain @xmath204 [ cf .
( [ eq : propagator - iii-8 ] ) ] .
similarly , we can also find that @xmath205 where @xmath206 represents an integral over the surface of a unit @xmath207-sphere ( with radius @xmath208 ) covering the first octant only .
along the same line , we continue to be able to obtain the corresponding expressions of @xmath209 and @xmath210 , respectively . in doing this job
, it can easily be induced that @xmath211 where @xmath212 .
now let us generalize the above scenario to the integral @xmath213 . applying the iterations to eqs .
( [ eq : propagator5 ] ) , ( [ eq : iterated1 ] ) and ( [ eq : iterated2 ] ) allows us to finally obtain the following irreducible expressions in terms of an integral over the higher - dimensional solid angle as @xmath214 and @xmath215 here the angle @xmath216 , and @xmath217 represents an integral over the first section only , corresponding to @xmath218 .
now we easily see that with the field strength @xmath219 , the integrals @xmath220 for @xmath221 , due to the fact that @xmath222 with @xmath223 . as a result ,
the quantity @xmath121 in ( [ eq : propagator - field1 - 1 - 0 - 0 ] ) can be rewritten in terms of the integrals over the higher - dimensional solid angle .
this result can also be interpreted in such a way that eq .
( [ eq : propagator - field1 - 1 - 0 ] ) for @xmath121 , given in form of the _ time - ordered _
integrals @xmath224 and so non - trivial to directly evaluate , is transformed , for an arbitrary time @xmath225 , into the integrals of some geometric pattern over the ( time - independent ) surface of a unit @xmath73-sphere , so being more accessible in numerical evaluation ; in fact , we see from ( [ eq : iterated1 - 1 ] ) and ( [ eq : iterated2 - 1 ] ) that the quantity @xmath213 is given by an irreducible @xmath226$]-dimensional integral , rather than an @xmath227-dimensional integral in ( [ eq : propagator - field1 - 1 - 0 ] ) , where the symbol @xmath228 $ ] is the greatest integer less than or equal to @xmath229 . due to this irreducible high dimensionality
, it is , apparently , highly non - trivial , though , to exactly evaluate the ( complicated ) integral @xmath213 for @xmath73 being large enough in terms of the hypergeometric functions @xmath184 simply with a single argument @xmath185 .
below we will accordingly explore an approximation scheme in which the quantity @xmath121 can be expressed in terms of the functions @xmath184 in reasonably simple form .
we already have the closed expression of the integral @xmath160 in terms of the bessel function in ( [ eq : propagator5 ] ) .
we next intend to evaluate the integral @xmath196 in ( [ eq : propagator7 ] ) in terms of some hypergeometric function . substituting first the identity @xcite @xmath230 with @xmath231 into ( [ eq : propagator7 ] )
, we can straightforwardly obtain @xmath232 where @xmath233 here the confluent hypergeometric function @xmath234 .
now we focus on explicit evaluation of the integral @xmath235 .
to do so , we first substitute into ( [ eq : propagator9 ] ) the expansion formula for a plane wave @xcite @xmath236 with @xmath237 and @xmath238 , where the bessel functions @xmath239 . as seen , each @xmath240 of the harmonic partial waves is then time - independent .
subsequently , with the help of another sum rule @xmath241 with @xmath242 followed by the beta function @xmath243 [ cf .
( [ eq : identity_gamma0 ] ) ] , we can easily obtain the integral representation @xcite @xmath244 in which @xmath245 . in doing so , we also used @xmath246 and @xmath247 @xcite .
( [ eq : identity_bessel1 ] ) and ( [ eq : propagator14 ] ) then allow us to have an evaluation of the integral @xmath235 and so @xmath248 this subsequently reduces to a compact expression @xmath249 in terms of the generalized hypergeometric functions @xmath250 ; \left[\frac{1}{2}-\frac{2l}{3 } , \frac{1}{2}+\frac{2l}{3 } , \frac{5}{6}-\frac{2l}{3 } , \frac{5}{6}+\frac{2l}{3 } , \frac{7}{6}-\frac{2l}{3 } , \frac{7}{6}+\frac{2l}{3}\right ] ; z\right)\,,\ ] ] which satisfies @xmath251 here we also used @xmath252 .
comments deserve here .
first , eq .
( [ eq : propagator17 ] ) can be rewritten as @xmath253 ; \left[\frac{5}{6 } , \frac{7}{6}\right ] ; \frac{i f^2 t^3}{32}\right)\ , + \right.\n\\ & & \left.\sum_{l=1}^{\infty } \frac{2\,i^{-l}}{1 - 16\,l^2}\ ; j_l\left(\frac{f^2 t^3}{64}\right)\ ; f_l^{(6)}\left(\frac{i f^2 t^3}{32}\right)\right\}\end{aligned}\ ] ] [ cf .
( [ eq : propagator22 ] ) and ( [ eq : bessel - function1 ] ) ] .
now it is explicitly shown that the quantity @xmath160 in ( [ eq : propagator5 ] ) is `` modulated '' by @xmath254 yielding in ( [ eq : propagator19 ] ) the partial wave @xmath255 as the leading term of @xmath196 , surrounded by the additional partial waves @xmath256 `` modulated '' by @xmath257 . secondly
, we may accordingly take the leading term only as a satisfactory approximation of @xmath196 ; in the weak - field regime ( @xmath258 ) leading to @xmath259 for @xmath260 with @xmath261 , we easily see the validity of this approximation .
it also applies sufficiently to the strong - field regime ( @xmath262 ) , in which due to the asymptotic behavior @xmath263 @xcite , the magnitude of all terms with @xmath264 can not be non - negligible enough .
in fact , we have @xmath265 anyway in this regime , as already pointed out after eq .
( [ eq : iterated2 - 1 ] ) . also , the strong - field regime corresponds to the semiclassical limit @xmath266 since the argument @xmath267 , expressed in a dimensionless unit , exactly corresponds to @xmath268 in the actual physical unit [ cf .
( [ eq : schroedinger_eq2 ] ) ] .
therefore the approximation with @xmath255 alone may also be considered an effective semiclassical treatment of @xmath196 .
next we explore a closed expression of @xmath170 in approximation . the substitution of the leading term of @xmath196 into eq .
( [ eq : propagator-7 - 0 ] ) allows us to straightforwardly obtain @xmath269 where @xmath270 here we applied the same technique as that used for ( [ eq : propagator8])-([eq : propagator9 ] ) .
subsequently we again apply to eq .
( [ eq : propagator - iii-2 ] ) both the partial wave expansion in ( [ eq : identity_bessel1 ] ) , followed by the selection of @xmath255 alone , and eq .
( [ eq : propagator14 ] ) with @xmath271 .
this immediately gives rise to @xmath272 substituting this into ( [ eq : propagator - iii-1 ] ) , we can next obtain straightforwardly @xmath273 where @xmath274 let @xmath275 .
then , this can easily be rewritten as @xmath276 here the second summation over the index @xmath73 precisely simplifies to @xmath277 ; \left[\frac{5}{6},\frac{7}{6},\frac{1}{2}-r\right ] ; -1\right)$ ] and can be approximated with satisfactory precision to its leading term @xmath278 alone , due to the fact that the summand @xmath279 for all @xmath280 ( cf . fig .
[ fig : fig1 ] ) . from this , eq .
( [ eq : propagator - iii-6 ] ) reduces to @xmath281 ; \left[\frac{2}{3},1,1,\frac{4}{3}\right ] ; \frac{i f^2t^3}{32}\right)\ ] ] which immediately gives rise to the expression @xmath282 ; \left[\frac{2}{3},1,1,\frac{4}{3}\right ] ; \frac{i f^2t^3}{32}\right)\ ] ] in terms of the generalized hypergeometric function .
as demonstrated in ( [ eq : propagator - iii-1])-([eq : propagator - iii-7 ] ) , without this leading - term approximation the quantity @xmath170 should be expressed in terms of a lengthy multiple sum , which is not transparent to physically interpret and not desirable for numerical evaluation . next the quantity
@xmath283 is under consideration .
we exactly follow the technique leading to eq .
( [ eq : propagator - iii-8 ] ) for @xmath170 ; we first substitute this previous result with ( [ eq : propagator - iii-7 ] ) into ( [ eq : propagator23 ] ) with @xmath284 and then carry out the same approximation , followed by applying eq . ( [ eq : propagator14 ] ) with @xmath285 . from this , we can arrive at the rather lengthy expression @xmath286 where @xmath287 [ cf .
( [ eq : propagator - iii-5 ] ) ] .
we now apply the technique used for ( [ eq : propagator - iii-6 ] ) and ( [ eq : propagator - iii-7 ] ) , which finally gives rise to @xmath288 ; \left[\frac{5}{6},\frac{7}{6},\frac{3}{2}\right ] ; \frac{i f^2t^3}{32}\right)\,,\ ] ] where @xmath289 ; \left[\frac{5}{6},\frac{7}{6},\frac{3}{2}\right ] ; \frac{i f^2t^3}{32}\right)\ ] ] [ cf .
( [ eq : propagator - iii-7 ] ) ] . along the same line
, we continue to be able to obtain the next expressions in approximation as @xmath290 ; \left[1,1,\frac{4}{3},\frac{5}{3}\right ] ; \frac{i f^2t^3}{32}\right)\\ & & \cdots\n\\ \mbox{i}_{11}(t ) & \approx & \frac{\{(2\pi i)^{-1}\ , t^9\}^{\frac{1}{2}}}{3\cdot 5 \cdot 7\cdot 9 i}\ , \exp\left(\frac{5 i f^2
t^3}{192 i}\right ) j_0\left(\frac{f^2 t^3}{64}\right)\ , { } _ { 4}\hspace*{-.07cm}f_4\left(\left[\frac{1}{2},\frac{5}{3},2,\frac{7}{3}\right ] ; \left[1,\frac{11}{6},\frac{13}{6},\frac{5}{2}\right ] ; \frac{i f^2t^3}{32}\right)\n\\ \mbox{i}_{12}(t ) & \approx & -\frac{t^5}{2 ^ 6\cdot 5!}\ , \exp\left(-\frac{5 i f^2 t^3}{192}\right ) j_0\left(\frac{f^2 t^3}{64}\right)\ , { } _ { 4}\hspace*{-.07cm}f_4\left(\left[\frac{1}{2},\frac{11}{6},\frac{13}{6},\frac{5}{2}\right ] ; \left[1,2,\frac{7}{3},\frac{8}{3}\right ] ; \frac{i f^2t^3}{32}\right)\n\\ & & \cdots \;.\n\end{aligned}\ ] ] by induction it easily follows that @xmath291 ; \left[1,\frac{n}{6},\frac{n+2}{6},\frac{n+4}{6}\right ] ; \frac{i f^2 t^3}{32}\right)\ ] ] where @xmath292 [ cf .
( [ eq : iterated1 - 1 ] ) and ( [ eq : iterated2 - 1 ] ) ] . here
the expression with @xmath293 is obviously meant as the leading term of eq .
( [ eq : propagator19 ] ) only .
it is also instructive to note that letting @xmath294 , eq .
( [ eq : semiclassial - n-1 ] ) directly recovers the exact expression of its field - free counterpart , explicitly given by @xmath295 , obtained from substitution of ( [ eq : kernel2 ] ) with @xmath66 into ( [ eq : propagator23 ] ) .
now we are ready to have a closed expression of the quantity @xmath121 in approximation , accounting for the ionization - recombination , as discussed after eqs .
( [ eq : exponential_decay_ansatz ] ) and ( [ eq : propagator - expand - field1 ] ) .
the substitution of ( [ eq : semiclassial - n-1 ] ) into ( [ eq : propagator - field1 - 1 - 0 - 0 ] ) explicitly reveals that @xmath296 ; \left[1,\frac{n}{6},\frac{n+2}{6},\frac{n+4}{6}\right ] ; \frac{i f^2 \tau'^3}{32}\right)\right\}\,.\end{aligned}\ ] ] this expression is useful especially in the weak - field limit , as stated after eq .
( [ eq : semiclassial - n-1 ] ) . to see transparently the strong - field behavior of @xmath297
, it may also be desirable to rewrite ( [ eq : propagator - field1 - 1 ] ) as follows ; we first decompose the sum index @xmath73 as @xmath298 , and @xmath299 , where @xmath300 .
next we plug the expansion @xmath301;[\cdots];z ) = \sum_{l=0}^{\infty } ( \cdots)\,z^l / l!$ ] into ( [ eq : propagator - field1 - 1 ] ) . with the aid of the sum identity @xmath302 where @xmath303 , we can unify and then simplify the two @xmath225-dependencies , explicitly given by @xmath304 and @xmath305 in the infinite sum over @xmath73 . this finally allows us to obtain @xmath297 as @xmath306 ; \left[\frac{3+j}{6},\frac{5+j}{6},\frac{7+j}{6},1-k,1-k\right ] ; \frac{i f^2 \tau'^3}{32}\right)\right\}\end{aligned}\ ] ] with @xmath219 [ note that the sum over @xmath307 is just a finite one ] . here
the generalized hypergeometric function @xmath308 can be understood as a compact form of @xmath309 , explicitly given by @xmath310;\right.\n\\ & & \left.\left[\frac{1-k}{2},\frac{1-k}{2},\frac{2-k}{2},\frac{2-k}{2},\frac{3+j}{12},\frac{5+j}{12},\frac{7+j}{12},\frac{9+j}{12},\frac{11+j}{12},\frac{13+j}{12}\right ] ; -\frac{f^4 \tau'^6}{2^{12}}\right)\,+\n\\ & & i\,\frac{f^2 \tau'^3}{32 } \left(\frac{1 - 2k}{2}\right ) \left(\frac{2+j}{6}\right ) \left(\frac{4+j}{6}\right ) \left(\frac{6+j}{6}\right)\,\times\n\\ & & { } _ { 9}\hspace*{-.07cm}f_{10}\left(\left[1,\frac{3 - 2k}{4},\frac{5 - 2k}{4},\frac{8+j}{12},\frac{10+j}{12},\frac{12+j}{12},\frac{14+j}{12},\frac{16+j}{12},\frac{18+j}{12}\right];\right.\n\\ & & \left.\left[\frac{1-k}{2},\frac{1-k}{2},\frac{2-k}{2},\frac{2-k}{2},\frac{3+j}{12},\frac{5+j}{12},\frac{7+j}{12},\frac{9+j}{12},\frac{11+j}{12},\frac{13+j}{12}\right ] ; -\frac{f^4 \tau'^6}{2^{12}}\right)\,.\end{aligned}\ ] ] and it is worthwhile to comment that although the function @xmath311 ; [ \cdots , 1-k,1-k ] ; \cdots)$ ] itself is not well - defined for @xmath312 , the composite expression @xmath311 ; [ \cdots , 1-k,1-k ] ; \cdots)/\{\gamma(1-k)\}^2 $ ] , given in ( [ eq : propagator - field2 - 1 ] ) indeed , is mathematically undisputed . as a result , the substitution of the quantity @xmath297 in ( [ eq : propagator - field1 - 1 ] ) or ( [ eq : propagator - field2 - 1 ] ) into ( [ eq : propagator - expand - field1 - 0 ] ) can directly yield an explicit expression of the propagator @xmath313 in terms of the generalized hypergeometric functions and an integral which can straightforwardly be evaluated to any sufficient degree of precision . here
we used @xmath162 and @xmath136 .
in fact , it can easily be shown that this remaining double integral is beyond the scope of its evaluation in terms of any generalized hypergeometric functions in reasonably simple form . from ( [ eq : propagator - expand - field1 ] ) , we can also obtain the corresponding wavefunction @xmath314 for an arbitrary initial state .
comments deserve here .
the wavefunction @xmath314 evolved from the initial bound state @xmath54 must , by construction , accommodate the ripples observed in the ionization probability @xmath315 in ( [ eq : ionization_prob1 ] ) . to have this feature
, we took the leading term only from each integral @xmath213 : in the strong - field limit , the first term @xmath316 on the right - hand side of eq .
( [ eq : propagator - expand - field1 ] ) for @xmath314 is dominant to the second term , given by a double integral , which has therefore been neglected in the analytical approach in @xcite . now , the result in ( [ eq : propagator - expand - field1 - 0 - 0 ] ) allows us to study systematically the next - order terms to @xmath316 ( in the intermediate - field regime ) , which are responsible indeed for the ripples resulting from the scattering from the delta - potential well at @xmath67 . in the weak - field limit , on the other hand , the second term ( corresponding to the influence of the residual zero - range potential ) is dominant to the first term subjected to the external field only , and so the quantity @xmath297 , and so @xmath317 , is a highly critical factor to determination of the entire time - evolution @xmath314 .
in fact , the bound - state amplitude @xmath318 of the ionization probability in eqs .
( [ eq : ionization_prob1])-([eq : homogeneous_ionization1 ] ) then reduces to a compact form as @xmath319 [ cf . ( [ eq : homog_sol1 ] ) for the closed expression of @xmath320
. finally , we also remark that in actual numerical evaluation of ( [ eq : propagator - expand - field1 - 0 - 0 ] ) we need to introduce a ( small ) constant @xmath321 in order to exactly fulfill the normalization condition @xmath322 in such a way that @xmath323 .
in summary , we have investigated the time evolution of a particle subjected to both a uniform electrostatic field and an attractive one - dimensional delta - function potential .
we have systematically derived the propagator @xmath80 of this system , in which its essential ingredient @xmath131 , accounting for the ionization - recombination in the bound - continuum transition , is exactly expressed in terms of the multiple hypergeometric functions .
and then , as our central finding , we have obtained the ingredient @xmath131 in ( appropriate ) approximation , expressed in terms of the generalized hypergeometric functions being much more transparent to physically interpret and much more accessible in their numerical evaluation than the multiple hypergeometric functions .
it has been shown that this finding provides a much better approximation scheme than the exponential decay approximation , where no ionization - recombination dynamics can be treated . in our approach , we have not applied the energy - time fourier nor the equivalent laplace transform , which has , on the other hand , been normally a starting point for analytical study of this system in references , e.g. , @xcite .
we think that our approach will therefore provide a useful framework for analytical study of the time - dependent ionization process in a delta - function potential well under an _ oscillatory _ electric field as the next task , for which one can not straightforwardly apply the convolution theorem of the laplace transform to the relevant lippmann - schwinger integral equation .
the author would like to thank g.j .
iafrate ( nc state university ) , y. wang , and l. zhou for stimulating discussions .
he also appreciates support from g.j .
iafrate during his stay with nc state university
. 1 t. yamabe , a. tachibana , and h.j .
silverstone , phys .
rev . a * 16 * , 877 ( 1977 ) , and references therein .
m. razavy , _ quantum theory of tunneling _ ( world scientific , singapore , 2003 ) . c. joachim and j.p .
launay , chem . phys . * 109 * , 93 ( 1986 ) .
m. jonson , phys .
b * 39 * , 5924 ( 1989 ) .
s. geltman , j. phys .
b * 10 * , 831 ( 1977 ) and j. phys .
b * 11 * , 3323 ( 1978 ) .
a. ludviksson , j. phys .
a * 20 * , 4733 ( 1987 ) .
w. elberfeld and m. kleber , z. phys .
b * 73 * , 23 ( 1988 ) . s.m . susskind , s.c .
cowley , and e.j .
valeo , phys .
a * 42 * , 3090 ( 1990 ) .
m. kleber , phys .
rep . * 236 * , 331 ( 1994 ) .
englert , lett .
34 * , 239 ( 1995 ) ; here , an `` analytical expression '' of the ionization probability @xmath324 was derived , which is , however , just a complicated formal expression in terms of some differential operator [ cf .
( 29 ) and ( 35 ) thereof ] .
a. rokhlenko and j.l .
lebowitz , j. math .
phys . * 41 * , 3511 ( 2000 ) ; here , the time - dependent wavefunction was expanded in terms of a complete set of eigenfunctions of the field - free hamitonian @xmath325 to study the system of a single delta - function potential well which is subject to _ time - dependent _ variations of the binding strength , but with no external field .
w. elberfeld and m. kleber , _ ionization and tunneling in a strong electric field _
, in _ physics of strong fields _ , edited by w. greiner ( plenum , new york , 1987 ) p. 465 .
reinhardt , _ time - dependent approach to spectra in external fields _ , in _ atomic excitation and recombination in external fields _ , edited by m.h .
nayfeh and c.w .
clark ( gordon and breach , new york , 1985 ) .
griffiths , _ introduction to quantum mechanics _
( pearson prentice hall , new jersey , 2005 ) .
damert , am .
j. phys . *
43*(6 ) , 531 ( 1975 ) .
o. valle and m. soares , _ airy functions and applications to physics _ , 2nd ed .
( world scientific , singapore , 2010 ) .
s. weinberg , _ the quantum theory of fields _ ( cambridge university press , 1995 ) . c. grosche
c and f. steiner , _ handbook of feynman path integrals _ ( springer , new york , 1998 ) .
m. moshinsky , phys .
88 * , 625 ( 1952 ) .
m. abramowitz and i. stegun , _ handbook of mathematical functions with formulas , graphs , and mathematical tables _
( dover , new york , 1974 ) . r.r .
schlicher , w. becker , j. bergou , and m.o .
scully , _ interaction hamiltonian in quantum optics or @xmath43 vs. @xmath326 revisited _ , in _ quantum electrodynamics and quantum optics _ , edited by a.o .
barut ( plenum , new york , 1984 ) p. 405 .
d. bauch , nuovo cimento b * 85 * , 118 ( 1985 ) .
w. elberfeld and m. kleber , am .
j. phys . * 56*(2 ) , 154 ( 1988 ) .
h. exton , _
multiple hypergeometric functions and applications _ ( halsted / wiley , new york , 1976 ) .
srivastava and p.w .
karlsson , _ multiple gaussian hypergeometric series _ ( halsted / wiley , new york , 1985 ) .
srivastava , j. phys .
a * 20 * , 847 ( 1987 ) .
sommerville , _ an introduction to the geometry of n dimensions _
( dover , new york , 1958 ) . eqs .
( [ eq : propagator14 ] ) , ( [ eq : propagator17 ] ) and ( [ eq : propagator18 ] ) as well as all following expressions in terms of the hypergeometric functions have been verified with the help of maple programs .
[ fig : fig1 ] : ( color online ) @xmath327 ; [ \frac{5}{6},\frac{7}{6},\frac{1}{2}-r ] ; -1)$ ] versus index @xmath328 with an interpolated line ( red dash ) .
this precisely represents the second summation over the index @xmath73 in eq .
( [ eq : propagator - iii-6 ] ) . in comparison ,
its approximation @xmath329 with an interpolated line ( blue dash ) , where @xmath330 .
this is simply the leading term @xmath278 of the summation . as seen , @xmath331 in a satisfactory manner .
|
we consider the time evolution of a particle subjected to both a uniform electrostatic field @xmath0 and a one - dimensional delta - function potential well .
we derive the propagator @xmath1 of this system , directly leading to the wavefunction @xmath2 , in which its essential ingredient @xmath3 , accounting for the ionization - recombination in the bound - continuum transition , is exactly expressed in terms of the multiple hypergeometric functions @xmath4 .
and then we obtain the ingredient @xmath3 in an appropriate approximation scheme , expressed in terms of the generalized hypergeometric functions @xmath5 being much more transparent to physically interpret and much more accessible in their numerical evaluation than the functions @xmath4 .
|
the observation of the colossal magnetoresistance effect ( cmr ) @xcite in doped manganese oxides with perovskite structure , r@xmath2a@xmath3mno@xmath4 ( r = rare - earth , a = alkaline - earth metal ) , moved these materials into the focus of intense research activity @xcite .
it turned out soon that the complex electronic and magnetic properties of the manganites depend on a close interplay of almost all degrees of freedom known in solid state physics , namely itinerant charges , localised spins , orbitals , and lattice vibrations . on the one hand ,
the strong coulomb interaction @xmath5 and the hund s rule coupling @xmath6 introduce a spin background and affect the charge mobility via double - exchange @xcite . on the other hand , the cubic environment of the mn sites within the perovskite lattice results in a crystal field splitting of mn-@xmath7-orbitals into @xmath8 and @xmath9 and
gives rise to an orbital degeneracy in the ground - state of mn@xmath10 ions .
this orbital degeneracy , in turn , connects the electronic system to the lattice , making it sensible to jahn - teller distortion and polaronic effects @xcite .
there are numerous attempts to describe the electronic ( i.e. , charge , spin , and orbital ) interactions of the manganites on a microscopic level ( see , e.g. , refs . )
. however , most of these models do not reflect the full multiplet structure of the mn ions caused by the cubic site symmetry or violate the spin rotational invariance . to our knowledge ,
feiner and ols @xcite were the first who derived a consistent spin - orbital model for the undoped compounds . in the following
we extend their derivation to the case of finite doping and complete the resulting electronic model with dynamic electron - lattice interactions of jahn - teller and breathing type .
up to now models of similar complexity have only been discussed on the basis of rather extensive approximations , where mean - field approaches are probably the most popular .
in particular , the lattice was always treated within the adiabatic approximation which is questionable in view of the comparable energy scales of spins and phonons . for a better understanding of this complex many body system
we prefer to consider all interactions on an equal footing . using exact diagonalisation techniques we study the ground - state properties of the derived model on a four site cluster . for the hole doped region ( @xmath1 )
our calculations show how the lattice can effectively control the spin and orbital correlations and the charge mobility . in the undoped compounds the electron - phonon interaction suppresses the coupling of spin and orbital degrees of freedom and is most effective in determining orbital order . at doping @xmath11 , where ferromagnetism is stabilised by the double - exchange mechanism ,
the coupling to the lattice can cause self - trapping of the charge carriers , which immediately switches the spin order . in turn , this change is reflected in dynamic lattice correlations , namely the bond length fluctuation .
for the half - doped manganites we find that the electron - phonon coupling enhances the susceptibility for charge ordering which is present already due to the strong coulomb interaction .
in addition , moderate variations in the strength of the electron - lattice coupling trigger pronounced changes in spin and orbital correlations .
this relates to the complicated patterns of spin , charge and orbital ordering observed experimentally for different manganites at @xmath12 .
the electronic and magnetic properties of the mixed - valence manganites are governed by the manganese @xmath7-electrons , which are divided into @xmath9 and @xmath8 according to the cubic symmetry within the crystal field . due to the large coulomb and hund s rule interactions each of the three @xmath9 and the two @xmath8 levels carries at most one electron and the spins of several @xmath7-electrons
are aligned in parallel at every site ( see figure [ figlevels ] ) .
starting from the undoped compounds with only mn@xmath10 ions doping will remove @xmath8-electrons .
hence , the local electronic hilbert space can be restricted to the large hund s rule ionic ground - states of manganese ions in a cubic crystal field ( see ref . ) . for mn@xmath10 ( @xmath13 )
this corresponds to the spin-@xmath14 orbital doublet @xmath15 [ @xmath16 , @xmath17 and for mn@xmath18 ( @xmath19 ) to the spin-@xmath20 orbital singlet @xmath21 [ @xmath22 , @xmath23 here the operators @xmath24 create spin @xmath25 electrons in the @xmath8 ( @xmath26 ) or @xmath9 ( @xmath27 ) orbitals and @xmath28 raises the on - site spin .
-electrons in a cubic environment .
here @xmath29 , @xmath6 , and @xmath30 denote the crystal field splitting , the hund s rule coupling , and the jahn - teller splitting , respectively . ] , eq . .
the shaded region corresponds to @xmath31 terms , the other terms are proportional to @xmath32 .
@xmath33 denotes different nonnegative prefactors of @xmath6 ( see eq . ) . ]
the overlap of neighbouring manganese @xmath7 and oxygen @xmath34 orbitals allows for a tunneling of the manganese electrons between adjacent sites . due to the specific symmetry of the involved orbitals
this hopping acquires an anisotropy @xcite , @xmath35 + \text{h.c.}\,,\end{gathered}\ ] ] which can be expressed in terms of the spatial rotations @xmath36 the transfer amplitudes @xmath37 and @xmath38 of @xmath8 and @xmath9 electrons are small compared to the local energies @xmath5 and @xmath6 .
therefore the mobility of the @xmath8 electrons depends on the correlations of the spin background formed by the @xmath9 electrons .
their kinetic energy is maximal , if neighbouring spins are aligned ferromagnetically , which is the essence of the well known double - exchange interaction @xcite .
second order processes in @xmath37 and @xmath38 are responsible for numerous superexchange interactions between the localised spins , which inherit the orbital anisotropy of the hopping . the electronic part of our model is derived by second order degenerate perturbation theory @xcite , i.e. , we calculated the matrix elements of @xmath39 between the basis states or and all admissible excited states @xcite . in figure [ figvirtual ]
the spin part of the fifteen different virtual excitations contributing to the superexchange is summarised graphically .
we neglect terms which involve three different lattice sites . a compact expression for the resulting electronic hamiltonian , @xmath40 is obtained by rewriting the orbital and charge degrees of freedom of the basis states ( [ eq5ebase ] ) and ( [ eqa2base ] ) in terms of new fermi operators @xmath41 and projectors @xmath42 , @xmath43 note that these projectors are related to the common pseudo - spin operators @xmath44 by the equation @xmath45 . since the amplitude of the on - site spin is different for the basis states ( [ eq5ebase ] ) and ( [ eqa2base ] ) , it is convenient to represent the spin degree of freedom by schwinger bosons @xmath46 and @xmath47 which allow for a uniform description .
then , the spin operator is defined by @xmath48 , but the amplitude is subject to the constraint @xmath49 . using this notation in @xmath50-direction the interaction between nearest - neighbour sites
is given by , @xmath51 + t_{\pi}^2\,\frac{\frac{4}{9}{\bf s}_{i}{\bf s}_{j}-1}{u+\frac{4}{3}j_h } \,p^{a_2}_{i}p^{a_2}_{j}\\ + t_{\pi}^2\,\frac{{\bf
s}_{i } { \bf s}_{j}-3}{3 } \left[\frac{(u-2j_h)(r_{x}(p^{\varepsilon}_{i}p^{a_2}_{j})+ r_{y}(p^{\varepsilon}_{i}p^{a_2}_{j } ) ) } { \frac{19}{3}j_h ( 2u-\frac{7}{3}j_h ) } + \frac{(u+\frac{5}{3}j_h)(r_{x}(p^{\theta}_{i}p^{a_2}_{j})+ r_{y}(p^{\theta}_{i}p^{a_2}_{j}))}{\frac{13}{3}j_h(2u - j_h ) } \right]\\ + t_{\pi}^2\,\frac{{\bf s}_{i } { \bf s}_{j}-4}{8 } \left[\frac{r_{x}(p^{\varepsilon}_{i}p^{\varepsilon}_{j})+ r_{y}(p^{\varepsilon}_{i}p^{\varepsilon}_{j})}{u+8j_h/3 } + \frac{r_{x}(p^{\theta}_{i}p^{\theta}_{j})+ r_{y}(p^{\theta}_{i}p^{\theta}_{j})}{u+2j_h } + \frac{(2u+\frac{14}{3}j_h)(r_{x}(p^{\varepsilon}_{i}p^{\theta}_{j})+ r_{y}(p^{\varepsilon}_{i}p^{\theta}_{j}))}{(u+4j_h)(u+\frac{2}{3}j_h ) } \right]\ , . \end{gathered}\ ] ] the first order in @xmath37 corresponds to the well known double exchange interaction @xcite , which the authors discussed in detail recently @xcite .
the second order in @xmath37 and @xmath38 appears to be more involved , since the number of accessible virtual excitations is rather large ( see figure [ figvirtual ] ) .
however , in all cases it is basically the product of a heisenberg - type spin interaction and two orbital projectors .
the energies of the virtual excitations depend , in general , on all three racah parameters @xcite @xmath52 , @xmath53 , and @xmath54 . applying the convention of ref .
we define @xmath5 and @xmath6 in terms of @xmath55 and @xmath56 excitations .
together with the approximation @xcite @xmath57 this yields @xmath58 and @xmath59 .
typical estimates for these energies are @xmath60 ev and @xmath61 ev
. @xcite concerning the electron - lattice interaction we concentrate on the local environment of the manganese ions , namely the surrounding oxygen octahedra ( see figure [ figmodes ] ) .
at every site two optical phonon modes of @xmath62 symmetry , @xmath63 and @xmath64 , couple to the orbital degree of freedom of the @xmath8 electrons .
in addition , a breathing - mode @xmath65 is sensitive to the electronic density .
to lowest order in the elongation @xmath66 ( @xmath67 ) this is modelled by the @xmath68 jahn - teller hamiltonian @xcite @xmath69\end{gathered}\ ] ] and a holstein - type @xcite interaction @xmath70 the bosonic operators @xmath71 describe the usual harmonic lattice dynamics . to a good approximation
we can assume these phonons to be dispersion - less , @xmath72 + \tilde\omega\sum_i b^{\dagger}_{i , a_1}b^{}_{i , a_1}\,.\ ] ] in experiments @xcite the vibrational modes involving distortions of the oxygen octahedra are found to have frequencies of the order of @xmath73 to @xmath74 mev . in the following
we assume @xmath75 and @xmath76 .
this approximation reduces the number of model parameters and is reasonable , since we are mainly interested in qualitative features of the electron - lattice interaction .
, @xmath64 ) and the breathing - type ( @xmath65 ) phonon modes . ]
the hilbert space of the complete microscopic model , @xmath77 is large and grows rapidly with the system size . however , using a density matrix based optimisation procedure @xcite for the phonon subsystem we are able to retain the full quantum dynamics of the lattice and the electronic subsystem in our numerical calculation @xcite of the ground - state properties .
in addition we take into account some of the discrete symmetries of the model , namely the conservation of the @xmath78 component of the spin , the particle number conservation , and the mirror symmetries orthogonal to the @xmath79 and @xmath80 axes ( dot - dashed in figure [ figcluster ] ) .
nevertheless , the typical dimension of the eigenvalue problem is of the order of @xmath81 and its repeated solution during the phonon optimisation requires the use of large scale computers . the undoped manganese oxides ( la@xmath82@xmath82mno@xmath4 , pr@xmath82@xmath82mno@xmath4 ) usually exhibit a - type anti - ferromagnetic order and strong jahn - teller distortion of the ideal perovskite structure @xcite . the origin of the observed magnetic order has been subject to discussions . while different band structure calculations @xcite emphasise the importance of lattice distortions for the stability of anti - ferromagnetism , feiner and ole @xcite favoured a purely electronic mechanism .
our calculation points out that both parameters , @xmath83 and @xmath84 , can drive a transition from ferromagnetic ( fm ) to anti - ferromagnetic ( afm ) order .
figure [ figd0order](c ) shows the phase diagram of the electronic model without electron - phonon interaction , i.e. , @xmath85 .
we assume @xmath86 ev and @xmath87 for the hopping integrals @xcite and characterise the magnetic phases according to the total spin @xmath88 of the ground - state of the four site cluster .
parameters in the range @xmath89 implicate that for two neighbouring sites a @xmath90 configuration becomes the ground - state in favour of the @xmath91 configuration .
this is incompatible with the situation in the manganites and consequently the electronic hamiltonian @xmath92 is not applicable for these values of @xmath5 and @xmath6 .
starting from the fm phase with @xmath93 both , increasing @xmath5 or @xmath84 , change the magnetic order of the ground - state to afm [ figure [ figd0order](a ) and ( b ) ] .
the magnetic transition is accompanied by a change in the corresponding orbital order , which we identify by the local expectation value @xmath94 and the orbital exchange correlation @xmath95 between neighbouring sites . here
the pauli matrices @xmath96 operate on the orbital degree of freedom , @xmath97 , at the site @xmath98 .
the panels ( a ) and ( b ) of the figures [ figd0order ] and [ figd0corr ] illustrate the transition from staggered to uniform orbital order . in view of the distinct driving interactions ( @xmath5 or @xmath84 )
both transitions appear to be very similar .
however , we observe a significant difference , if we study the ( de)coupling of spin and orbital degrees of freedom . the latter has been a rather controversial issue in the case of the kugel - khomskii model @xcite , @xmath99\,,\end{gathered}\ ] ] which contains the same kind of spin - orbital interactions @xmath100 as our hamiltonian .
using equation of motion approaches for the approximate solution of the model khaliullin and oudovenko @xcite decoupled spin and orbital degrees of freedom , while feiner , ols and zaanen @xcite emphasised the role of mixed spin - and - orbital excitations .
a correlation function , which makes it possible to distinguish between these two decoupling schemes , is the spin - orbital fluctuation @xmath101 .
the data we calculated for the complete manganite model is given in figure [ figd0corr](c ) and ( d ) .
it indicates that this correlation is smaller by a factor of @xmath102 to @xmath103 , if phonons are responsible for the fm to afm transition ( open symbols ) . on the other hand ,
starting within the afm phase increasing electron - phonon interaction clearly suppresses the spin - orbital - fluctuation [ figure [ figd0corr](c ) , bold symbols ] .
this behaviour is , of course , crucial for effective theories that are based on such decoupling schemes . at doping @xmath104 .
] at doping @xmath104 . ] at doping @xmath104 . ]
@xmath105 to characterise the behaviour of the lattice we calculate the correlations @xmath106 ( @xmath107 ) between the elongations of the jahn - teller modes @xmath63 and @xmath64 at neighbouring sites @xmath98 , @xmath108 .
the bond lengths @xmath109 define additional significant quantities .
in particular we considered the expectation values @xmath110 and @xmath111 with @xmath112 .
figure [ figd0latt ] shows that within the fm domain the lattice is undistorted but the @xmath64 vibrations are correlated along the @xmath113 direction in reciprocal space . at intermediate values of @xmath114 a finite @xmath79-@xmath80-distortion develops , which is also reflected in spin and orbital correlations [ e.g. , the spin - orbital fluctuation in figure [ figd0corr](c ) ] .
such type of asymmetry seems surprising in view of the invariance of the cluster with respect to diagonal reflections ( dot - dot - dashed in figure [ figcluster ] ) .
however , since this symmetry is not taken into account in the calculation the system is trapped in one particular linear combination of two degenerate ground - states and different expectation values in @xmath79 and @xmath80 direction can evolve . at larger electron - phonon coupling @xmath84
only a finite @xmath63 distortion remains and the @xmath64 modes are uncorrelated . in figure [ figorbph0 ]
the change of orbital , spin and phonon correlations with increasing @xmath84 is shown schematically .
this , however , is a rather suggestive picture , which can not cover all the details of the various correlations . to select the appropriate graphical representation for the orbital arrangement we study the @xmath115-eigenstates of reduced orbital density matrices . on a bond @xmath116
these states are either anti - symmetric , @xmath117 or symmetric , @xmath118 with respect to the permutation of two sites .
the angles @xmath119 parameterise two generalised orbital states of the form @xmath120 . at small @xmath84
the staggered orbital order implies that the anti - symmetric orbital state is the most probable for each bond , i.e. , it belongs to the largest eigenvalue of the density matrix . since an anti - symmetric combination of two arbitrary generalised states @xmath121 and @xmath122 is always proportional to @xmath123 the representation is undefined .
therefore we took into account also the second eigenstate of the density matrix , which is slightly less probable but necessarily symmetric . the associated angles @xmath124 and @xmath125 define the orbital pattern shown in the left hand panel of figure [ figorbph0 ] . for larger @xmath84
the uniform orbital order allows for a calculation of the depicted orbital pattern directly from the most probable eigenstate of the density matrix .
the spin arrangements are chosen such as to reflect the expectation values of the heisenberg interaction , @xmath126 , for nearest neighbour bonds @xmath116 .
of course , the calculated ordering patterns can not be directly compared to experimental observations for the three dimensional compounds .
nevertheless , the orbital arrangement within the fm spin background closely resembles the data measured within the fm planes of the a - type afm structure @xcite . at doping @xmath127 .
] @xmath128 at doping @xmath127 . ] in view of the cmr effect the domain of low to intermediate doping ( @xmath129 ) is certainly the most interesting one . here
ferromagnetism is stabilised by the double exchange interaction , which depends on the mobility of the charge carriers . however , if too strong electron - phonon coupling causes localisation of the holes , the spin order breaks down . for @xmath11
our calculation of the ground - state properties clearly illustrates this coincidence .
figure [ figd1all](a ) shows the total spin @xmath88 of the cluster together with the expectation value of the kinetic energy in the ground state , @xmath130 + \text{h.c.}\right\rangle\,.\ ] ] obviously , the change from fm to afm spin correlations with increasing electron - phonon interaction @xmath84 is directly related to the charge mobility .
in addition the transition is accompanied by the appearance of a lattice distortion in the @xmath79-@xmath80-plane [ see figure [ figd1all](c ) ] , which is interesting also in view of the lattice dynamics . whereas one of the elongations , @xmath131 , grows linearly in @xmath84 , the associated fluctuation @xmath132 shows a kink near the fm to afm transition point .
this behaviour closely reminds the data collected by booth et al .
@xcite using x - ray - absorption fine - structure measurements ( xafs ) . for la@xmath2ca@xmath3mno@xmath4
these authors observed a similar rise in the mn - o bond length variance @xmath133 near the critical temperature of the transition from the ferromagnetic metallic to the paramagnetic insulating phase .
orbital correlations are again extracted from the eigenstates of the reduced density matrices for a single bond .
we concentrate on the environment of a hole , which is characterised by the @xmath134 eigenstates , i.e. , by states describing a mn@xmath10 site neighbouring a mn@xmath18 site .
each of these states can be understood as the product @xmath135 of a generalised orbital state @xmath121 and the basis state @xmath136 [ see eq . ] . in figure
[ figd1all](b ) the corresponding angle @xmath124 of the most probable eigenstate of the density matrix is given for bonds in @xmath79 and @xmath80 direction .
the data for weak and strong electron - phonon coupling @xmath84 is translated into the orbital arrangement presented in figure [ figorbph1 ] .
increasing @xmath84 reduces nearest - neighbour correlations and forces the orbital order to depend locally on the dominant electron - phonon interaction .
the orbital polaron @xcite pattern disappears , if the charge carrier is trapped completely . at doping @xmath12 .
] at doping @xmath12 . ] at doping @xmath12 . ]
@xmath105 at doping @xmath12 the picture is more involved .
again we consider model parameters , which yield ferromagnetic spin correlations for vanishing electron - phonon coupling , @xmath85 .
the strong coulomb interaction @xmath5 then leads to the formation of a charge density wave with ordering vector @xmath113 in reciprocal space .
increasing electron - phonon coupling @xmath84 amplifies this trend , and at large @xmath84 the trapping of the carriers gives rise to charge order . in figure
[ figd2all](c ) this is illustrated with the density - density correlation between different lattice sites @xmath98 and @xmath108 . from the model
hamiltonian it is obvious that if charges tend to maximise their mutual distance the particular anti - ferromagnetic component of the heisenberg spin interaction , which is proportional to @xmath137 , gains importance .
consequently fm order is unstable at much lower @xmath84 , if compared to the case @xmath11 .
the fm to afm transition is not connected to charge localisation and causes only a tiny jump of the kinetic energy [ see fig .
[ figd2all](a ) ] .
the calculated orbital correlations are interesting as well . on a diagonal bond the most probable @xmath115-eigenstate of the reduced orbital density matrix is symmetric for
all considered values of @xmath84 , @xmath138 .
note however , that the associated angles are _ complex _ conjugated @xmath139 within both the fm and parts of the afm phase [ see figure [ figd2all](b ) ] .
here exact diagonalisation supports recent mean - field studies @xcite , which obtain different patterns of complex orbital ordering , in particular within the ferromagnetic conducting phase . at large @xmath84 charge localisation
promotes real orbital states . in figure [ figorbph2 ]
we summarise the behaviour of the lattice , the spins and the orbitals . with growing @xmath84
the cluster expands isotropically until a finite @xmath79-@xmath80-distortion develops at the fm to afm transition [ see also figure [ figd2all](d ) ] .
further increase of @xmath84 causes the carriers to localise and the distortion to disappear .
the spin background undergoes a change from fm order to different types of afm order .
of course , within the four site system we are not able to detect more complicated spin arrangements like the ce - type order observed experimentally @xcite .
however , the sensitivity of the system to small changes in the model parameters is clearly visible .
note that the depicted orbitals represent the amplitude of the underlying complex states @xmath121 .
in the present work we have derived a microscopic model for doped cmr manganites which includes the dynamics of charge , spin , orbital , and lattice degrees of freedom on a quantum mechanical level . using exact diagonalisation techniques we have studied how the electron - lattice interaction affects short range spin and orbital correlations .
an observation , which is important for the understanding of the undoped compounds , is the suppression of the spin - orbital coupling with increasing electron - phonon interaction . for the weakly doped compounds we have demonstrated the direct relationship between the trapping of charge carriers and the breakdown of ferromagnetism . in addition , we have shown that changes in the spin correlations are reflected in dynamic lattice correlations . at intermediate doping
we find that the system depends on a subtle balance of double - exchange , superexchange and electron - lattice interaction .
the latter enhances the tendency for charge ordering which in turn affects spin and orbital order . besides the calculation proves that complex orbital states can be a suitable approximation for the description of orbital correlations .
although the calculated data is not quantitatively comparable to real , three dimensional materials , the exact diagonalisation of even a small system provides detailed insight into correlations and driving interactions behind the rich phase diagram of the manganites .
in addition , the exact results may support the development of approximate theories .
we thank l.f .
feiner , f. ghmann , d. ihle , g. khaliullin , and j. loos for valuable discussions .
the grant of computational resources by hlr stuttgart and nic jlich , as well as financial support by the deutsche forschungsgemeinschaft and the czech academy of sciences is acknowledged .
|
we derive a realistic microscopic model for doped colossal magnetoresistance manganites , which includes the dynamics of charge , spin , orbital and lattice degrees of freedom on a quantum mechanical level .
the model respects the @xmath0 spin symmetry and the full multiplet structure of the manganese ions within the cubic lattice . concentrating on the hole doped domain ( @xmath1 ) we study the influence of the electron - lattice interaction on spin and orbital correlations by means of exact diagonalisation techniques .
we find that the lattice can cause a considerable suppression of the coupling between spin and orbital degrees of freedom and show how changes in the magnetic correlations are reflected in dynamic phonon correlations .
in addition , our calculation gives detailed insights into orbital correlations and demonstrates the possibility of complex orbital states .
|
the cooling of neutron stars offers a unique diagnostic of the physics of the interior .
because the emergent spectrum is nearly blackbody , the observed flux can be modeled to estimate the temperature , the intervening column of absorbing material , and the ratio of radius to distance .
astrometric measurement of the distance then allows an estimation of the photospheric radius , which is sensitively dependent on the high density equation of state .
psr b0656 + 14 is a middle - aged ( @xmath5 yr old ) pulsar , with a spectrum dominated at ultraviolet and soft - x - ray wavelengths by a @xmath6 k blackbody component , as expected in standard cooling models .
distance estimates obtained from measurements of radio dispersion in the ionized material between earth and the pulsar have led to a large estimated stellar radius , @xmath7 km . here
we show that the parallax of psr b0656 + 14 , determined using very long baseline interferometry , leads to a factor three reduction in the estimated pulsar distance .
we place this result in the context of existing thermal models for the neutron star .
for many years , pulsar parallax measurements have been just at or beyond experimental limits .
the situation has greatly improved with the increased sensitivity available using the nrao very long baseline array ( vlba ) , together with new phase - referencing techniques for measuring and eliminating ionospheric distortion and the ability to gate the vlba correlator to accept data only during intervals when the pulsar is `` on . '' in the last three years , parallaxes ( and distances ) to ten pulsars have been measured with the vlba ( brisken et al .
2002 ; chatterjee et al .
2001 ) .
psr b0656 + 14 has a 1.4 ghz flux density of 3.6 mjy , which is too weak to use the ionosphere calibration technique described by brisken et al .
( 2002 ) . instead , the in - beam calibration technique used by chatterjee et al .
( 2001 ) was applied .
very large array ( vla ) observations were made in the a - configuration at 1.4 ghz to search for potential in - beam calibrators .
three candidates were followed up at 8.4 and 15 ghz to test for compactness .
fortunately one of them , 0658 + 1410 , was compact and bright enough for use as an in - beam calibrator .
see table [ tab : calibrators ] for a list of calibrators used , their coordinates , and their flux densities .
five observations of psr b0656 + 14 , each five hours long , were made over 1.3 yrs with the vlba .
a 32 mhz band centered at 1.6675 ghz was used , resulting in a beam size of @xmath8 mas .
each observation consisted of sixty 2 minute scans on the field containing the pulsar and in - beam calibrator , each bracketted by observations of j0637 + 1458 , and two 5 minute scans on 3c147 for bandpass calibration .
the target field was correlated twice .
the first correlation placed the in - beam calibrator at the phase center .
the second put the pulsar at the phase center and used the pulsar gate to improve the signal - to - noise .
considerable effort went into ensuring that each observation was reduced identically .
analysis used the standard aips package .
after editing the data to remove radio - frequency interference , data from all five epochs were combined to make a single model for each calibrator source using standard techniques .
after initial amplitude and bandpass calibration , gps - based ionosphere calibrations were applied with the task tecor .
j0637 + 1458 was then fringe fitted ( with the model ) to determine the delays and rates and to perform a first - order phase calibration , which was followed by phase - and - amplitude self - calibration .
calibration derived from j0637 + 1458 was interpolated in time and applied to 0658 + 1410 .
this source was self - calibrated for phases only , again with a model .
the calibrations from both j0637 + 1458 and 0658 + 1410 were applied to the pulsar , images were produced , and image - plane gaussian fitting performed to determine the pulsar position
. uncertainties are dominated by systematic effects , mainly due to atmospheric and ionospheric refraction . to estimate the size of these effects
, imaging was repeated at each epoch with subsets of the data .
the measured position varied most with varying time cuts .
the scatter in position when varying the time ranges was used to assign error bars .
final single - epoch position uncertainties were 0.8 mas and 0.4 mas in right ascension and declination respectively .
the position of b0656 + 14 at each epoch is shown in table [ tab : positions ] .
note that the absolute positions are accurate only to about 10 mas due to the uncertainty in the true position of the calibrator 0658 + 1410 .
the five measured positions were used to simultaneously fit the parallax , @xmath9 , proper motion ( @xmath10 and @xmath11 ) , and position of psr b0656 + 14 .
these values and the derived distance , transverse velocity , @xmath12 , and line - of - sight column density of electrons , @xmath13 are shown in table [ tab : derived ] .
figure [ fig : pi ] shows the parallax fit to the data points . in this plot ,
the proper motion has been removed from the data to allow the subtle parallax to be seen .
the proper motion measured here differs from that measured by thompson and crdova ( 1994 ) , @xmath14 and @xmath15 mas / yr , by more than @xmath16 .
it is in excellent agreement with the _
hst_-derived value , @xmath17 and @xmath18 mas / yr ( mignani , de luca , & caraveo 2000 ) , though somewhat more precise .
the distance determined from the pulsar dispersion measure ( dm ) using the galactic electron distribution model of taylor and cordes ( 1993 ) is @xmath19 pc . using the electron distribution model of cordes and lazio ( 2002 ) yields @xmath20 pc .
both values are substantially greater than the parallax distance , @xmath1 pc .
the discrepancy is not entirely surprising the galactic electron density is poorly sampled at this @xmath21 due to a lack of calibration sources near the @xmath22 galactic latitude of psr b0656 + 14 ( anderson et al .
furthermore , the nearest calibration source for the taylor & cordes model , psr b0823 + 26 at @xmath23 , is known to have a dm derived distance some 3 times its measured parallax ( gwinn et al .
1986 ) . nevertheless , the high mean electron density along this path , @xmath24 @xmath25 , is a reminder that dispersion distances are particularly uncertain in the solar neighborhood , where lines of sight do not average over many independent regions .
psr b0656 + 14 has been extensively studied at optical ( shearer et al.1997 ; kurt et al .
1998 ; koptsevich et al .
2001 ) , ultraviolet ( pavlov , stringfellow , & crdova 1996 ; pavlov , welty , & crdova 1997 ; edelstein et al .
2000 ) , x - ray ( crdova et al . 1989 ; finley , gelman , & kizilolu 1992 ; anderson et al . 1993 ; greiveldinger et al . 1996 ; mineo et al .
2002 ; marshall & schulz 2002 ; pavlov et al .
2002 ) , and gamma - ray ( ramanamurthy et al . 1996 ) energies .
a simple model that adequately accounts for observations from the optical through the x - ray bands includes a power law component , identified with non - thermal emission from the active magnetosphere , and two blackbody components : a harder component identified with the heated polar caps , and a softer component identified with emission from the entire photosphere . in general
, model fits agree very well .
for example , a model based on the _ euve , rosat , _ and _ asca _ data ( koptsevich et al .
2001 ) gives @xmath26 k for the temperature of the soft component , while an independent model based on _ chandra _ observations ( marshall & schulz 2002 ) gives @xmath27 k , where @xmath28 is the gravitationally redshifted temperature measured by a distant observer : @xmath29 .
for our distance of 288 pc , the resulting estimates for the radius of the neutron star , @xmath30 are 7.8 and 8.5 km , respectively . including the optical / uv data and forcing a single power - law slope for the non - thermal component across more than four orders of magnitude in energy , pavlov et al .
( 2002 ) find a slightly smaller radius , 6.9 km at 288 pc .
estimates for the radius of the hot polar cap vary from 0.5 to 0.64 km at that distance . a blackbody radius
as small as @xmath31 km would pose serious challenges to standard neutron star models ( e.g. , heiselberg & pandharipande 2000 ) , though it would be acceptable for a low mass strange star ( alcock , fahri , & olinto 1986 ) .
however , a more likely explanation for this anomalous result is failure of the simple black - body modeling . in general ,
realistic atmosphere models tend to shift flux into high energy tails that result in overestimation of the effective temperature and hence underestimation of the stellar radius ( see romani 1987 and shibanov et al .
1992 for early work on non - magnetic and magnetic atmospheres , respectively , and zavlin & pavlov 2002 for a recent review ) .
the largest radii are found with non - magnetic hydrogen or helium atmospheres , while strong magnetic fields suppress the high energy tails and produce intermediate temperature and radius estimates .
our blackbody model radius should therefore most properly be treated as a lower limit to the neutron star radius . for psr b0656 + 14
, fits to unmagnetized hydrogen atmospheres ( e.g. , marshall & schulz 2002 ) produce very large radius estimates : @xmath32 km at our measured distance .
this can be regarded as an ( unconstraining ) upper limit to the radius .
magnetized atmospheres produce much more reasonable results . using just the _ rosat _ data , the magnetic hydrogen models of shibunov et al .
( 1993 ) , and a canonical neutron star radius of @xmath33 km , anderson et al .
( 1993 ) estimated the pulsar distance to be @xmath34 pc , in remarkable agreement with our measurement .
scaling from model fits done for the geminga pulsar , marshall & schulz ( 2002 ) estimate that the using a magnetized hydrogen atmosphere will increase the radius by a factor 2.3 over a blackbody estimate , resulting in corrected radius estimates of 15.9 , 17.9 , and 19.7 km for the pavlov et al .
( 2002 ) , koptsevich et al.(2001 ) , and marshall & schulz ( 2002 ) model fits , respectively .
we conclude that within the model uncertainties , any radius @xmath35 between @xmath3 and @xmath4 km is allowed .
a promising observational approach is to complement the high - energy observations with improved constraints on the flux in the optical and ultraviolet bands .
anderson et al .
( 1993 ) found that the blackbody and magnetized hydrogen atmosphere models that each fit the _ rosat _ data have significantly different predictions in the optical bands : for example , @xmath36 and @xmath37 , respectively . analysis of the combined integrated photometric data indicate that the spectral energy distribution is predominantly thermal in the ultraviolet , then flattens out and asymptotically approaches a power - law - like form towards the infrared ( koptsevich et al .
the photometry is well fit by a rayleigh - jeans plus synchrotron - like power - law model
the former agreeing to first order with the x - ray extrapolations .
the power - law , with a best - fit index @xmath38 , is similar to that observed at hard - x - ray and gamma - ray energies . in the ultraviolet ,
comparisons with model predictions are challenging because of the uncertain reddening . in the optical ,
the challenge has been to detect the thermal contribution beneath the larger non - thermal power - law .
the observed @xmath39 magnitude is @xmath40 ( kurt et al .
1998 ) , more than an order of magnitude above the expected thermal contribution . of considerable interest , then , has been the detection of optical pulsations .
shearer et al .
( 1997 ) reported pulsations based on @xmath39-band observations with a 2-d counting mama camera , and kern ( 2002 ) detected pulsations in the @xmath41 band using a phase - binning ccd system .
in addition to confirming the magnetospheric origin of the optical radiation , these observations raise the possibility of measuring and removing the non - thermal component . while there are clear differences in the light curve morphologies , both datasets are consistent with the radiation being 100% pulsed . at this time
, only upper limits can be placed on any unpulsed thermal emission : in @xmath39 , the limit is @xmath42 .
this is within a factor @xmath43 of the predicted value . of considerable interest
would be extending these observations to @xmath44 band , where the nonthermal flux will be smaller and the predicted thermal flux is a factor four higher . these observations will be challenging , as will continued improvement in the magnetic atmosphere modeling .
but both observations and theory have seen remarkable improvement over the last decade , even when the ultimate limit was the very uncertain pulsar distance .
now the distance to psr b0656 + 14 is known to 10% , comparable to the typical uncertainty of individual spectral model fits , and somewhat below the systematic uncertainties in the atmospheric fitting .
there is no reason that continued radio observations should nt continue to improve the parallax limit : precision a factor five better has now been achieved for psr b0950 + 08 .
thus there is every reason to believe that psr b0656 + 14 will continue to develop into a fascinating astrophysical laboratory for nuclear physics .
we wish to thank andrew lyne for providing us with the pulsar ephemeris that allowed us to gate the vlba correlator .
s.e.t . is supported by the nsf under grant ast-0098343 .
is supported by enterprise ireland under grant sc/2001/322 .
pavlov , g. g. , zavlin , v. e. , & sanwal , d. 2002 , in neutron stars , pulsars , and supernova remnants , ed .
w. becker , h. lesch , & j. trmper ( garching bei mnchen : max - planck - institut fr extraterrestrische physik ) , 273 .
zavlin , v. e. & pavlov , g. g. 2002 , in neutron stars , pulsars , and supernova remnants , ed .
w. becker , h. lesch , & j. trmper ( garching bei mnchen : max - planck - institut fr extraterrestrische physik ) , 263 lllrrl 3c147 & @xmath45 & @xmath46 & 1493 & 2998 & bandpass calibrator + j0637 + 1458 & @xmath47 & @xmath48 & 234 & 250 & phase calibrator + 0658 + 1410 & @xmath49 & @xmath50 & 33 & 38 & in - beam calibrator + ll @xmath61 & @xmath62 + @xmath63 & @xmath64 + @xmath10 & @xmath65 mas yr@xmath66 + @xmath11 & @xmath67 mas yr@xmath66 + @xmath9 & @xmath68 mas + distance & @xmath1 pc + @xmath12 & @xmath69 km s@xmath66 + @xmath13 & @xmath70 @xmath25 +
|
we present the result of astrometric observations of the radio pulsar psr b0656 + 14 , made using the very long baseline array .
the parallax of the pulsar is @xmath0 mas , yielding a distance @xmath1 pc .
this independent distance estimate has been used to constrain existing models of thermal x - ray emission from the neutron star s photosphere .
simple blackbody fits to the x - ray data formally yield a neutron star radius @xmath2 km . with more realistic fits to a magnetized hydrogen atmosphere
, any radius between @xmath3 and @xmath4 km is allowed .
|
entropy is a measure of disorder or randomness , and assumes its maximal value when a system can be in a number of states randomly with equal probability , and is minimally zero when the system is in a given state , with no uncertainty in its description .
apart from this common feature shared by all definitions of entropy at two ends of the scale , variations are possible in particularizing the functional form in between @xcite .
they lead to different forms of the probability distributions for states with different energies or some other conserved attribute .
some turn up as extensive , where the entropy of a combination of systems is simply the sum of the entropies of the systems , as in the classical case of the shannon form , while others can be defined to be not so .
renyi entropy @xcite is different from shannon , yet extensive , and hence the shannon form is not unique with respect to the property of extensivity .
tsallis entropy @xcite has attracted a lot of attention in recent years , not only on account of its conceptual and theoretical novelty , but also because it can be shown in specific physical cases @xcite to be the relevant form where nonextensivity is expected on account of the interaction of the combined subsystems . in the proper limiting case it reduces to the standard shannon entropy , indicating the consistency of the concept . in this paper , however , we shall introduce entropy from a new perspective , which too will bear semblance to the normal form in the limit .
we shall first present the rationale for this new definition and compare it briefly with the forms already being used .
then we shall find the form of the probability distribution for this entropy , which we shall henceforth call s - entropy , as it will be seen to be related to the concept of rescaling of the phase space .
let us consider a register of only one letter .
let @xmath0 be the set of probabilities for each of the @xmath1 letters @xmath2 that can occupy this position .
we are here using the language of information theory , as used for example , in the shannon coding theorem , though it is trivially extensible to states @xmath3 of a single state of an ensemble where the individual systems can be in any @xmath1 states with probabilities @xmath4 .
let us now consider a small deformation of the register to a new size so that it can accommodate @xmath5 letters .
the probability that the whole new phase space is occupied by the letter @xmath4 is now @xmath6 by the corresponding and operation and hence the probability that the new deformed cell is occupied by any of the pure letters @xmath2 is @xmath7 for @xmath8 this would give a shortfall from the original total probability of unity for @xmath9 .
it is obvious that the shortfall , which we denote by @xmath10 represents the total probability that the mixed cell has a mixture of @xmath2 and some other @xmath11 fractionally , since the total probability that the cell is occupied by one or more ( fractional included ) letters must be unity .
hence the mixing probability @xmath12 is actually a measure of the disorder introduced by increasing the cell scale from unity to @xmath13 .
the introduction of fractional values of cell numbers can be taken in the same spirit as defining the fractal ( hausdorff ) dimensions of curves , and in complex systems there have been studies of diffusion @xcite and percolation in complex systems with effectively fractional dimensions for fluids where the special geometric constraints translate into a change in the dimension of the corresponding space to an apparently nonintuitive fractional dimension . in coding theories for optimal transmission of information @xcite , we come across huffmann coding , where the optimum alphabet size may be formally a fraction , though for practical purposes it may be changed to the nearest higher integer . in probabilistic optimization , we may therefore consider a fractional size of the registrar , or equivalently , an integral number of cells in the registrar with fractional sized cells to accommodate a given amount of information .
probabilistic optimization in place of the deterministic parameterization of classical shannon information theory @xcite becomes inevitable in quantum computing contexts , and hence our use of the fractional cell sizes may be a classical precursor of the inevitable departure from stringent shannon - type concepts .
for an alphabet of @xmath14letters we define the entropy from the information content of the registrar by @xmath15 so that the entropy indicates an effective change in the mixing probability due to an infinitesimal change in the cell - size of the registrar .
this leads to @xmath16 in other words @xmath17 * we have some material at the other place .
* this differential form is analogous to but different from the tsallis form @xmath18 where there is an apparent singularity at @xmath9 which is the shannon limit .
the difference between the tsallis expression and ours becomes clearer if we express entropy as the expectation value of the ( generalized or ordinary ) logarithm .
@xmath19 where the generalized q - logarithm is defined as @xmath20 the expectation value is defined in terms of the simple probability distribution @xmath21 in our case we define the expectation value with respect to the deformed probability corresponding to the extended cell , while keeping the usual logarithm @xmath22 with @xmath23 in the limit @xmath24 @xmath25 approaches the normal logarithm , and hence tsallis entropy coincides with shannon entropy and also as @xmath26 we too get the normal shannon entropy .
the renyi entropy is defined by @xmath27 like shannon entropy this one is also extensive , i.e. simply additive for two subsystems for any value of @xmath28 . to get shannon entropy
uniquely one needs @xcite a slightly different formulation of the extensivity axiom @xmath29 where @xmath30 is the entropy of subsystem @xmath31 given subsystem @xmath32 is in state @xmath3 .
the @xmath4 can be obtained in terms of the energy of the states , or possibly also other criteria in the usual way by maximizing the entropy with constraints @xmath33 and @xmath34 .
the solution of the optimization equation gives for energy @xmath35 the probability @xmath4 @xmath36 where @xmath37 and w(z ) is the lambert function defined by @xmath38 here @xmath39 and @xmath40 are constants coming from the lagrange s multipliers for the two constraints and are related to the overall normalization and to the relative scale of energy , i.e. to temperature ( @xmath41 ) as in the shannon case where we get the gibbs expression for @xmath4 . in the tsallis case
@xmath4 has the well - known value @xmath42 which is easily seen to reduce to shannon form for @xmath43 . after some algebra it can be shown that this form reduces to the shannon form for @xmath24 .
the nonextensivity of tsallis entropy is seen easily by expanding @xmath44 for renyi entropy we have the simple additive relation @xmath45 in case of the new entropy @xmath46 where the @xmath47 are the mixing probability of states for subsystem @xmath39 as defined in eq.2 .
in fig.1 we show the variation of the probability function for different @xmath48 at different @xmath28 values .
comparison of the pdf for the new entropy for values of @xmath49 and @xmath50 .
the solid line is for @xmath9 , i.e. the gibbs exponential distribution and the lines are in the order of @xmath28,width=302 ] we note that the pdf drops increasingly rapidly for higher values of @xmath28 , and is quite different in shape and in magnitude at high energy values from the gibbs exponential distribution .
a variation of even @xmath51 from the standard value of @xmath9 can cause a quite discernible change in the pdf and should be observable in experimental contexts fairly easily . at @xmath52 ,
the shape is almost linear . in fig.2 and fig.3
we show the comparison of tsallis pdf and the pdf for the new entropy for the same values of @xmath28 , 1.1 in the former and 1.3 in the latter .
we notice that for larger @xmath28 values the new entropy gives much stiffer probability functions departing substantially from the tsallis pdf s .
comparison of pdf s for tsallis nonextensive entropy and the new entropy presented here , for @xmath53,width=302 ] the same as fig . 2 , but for a higher @xmath52,width=302 ]
we see that the new entropy presented here based on the simple concept of the amount of mixing of states freedom introduced per unit cell of phase space leads to a nonextensive form different from any of the presently studied entropies .
it leads to a complicated , but still integrable form of the pdf which departs substantially from tsallis entropy .
this entropy is also nonextensive in a fashion different from tsallis entropy , though like tsallis it too becomes extensive trivially in the limit @xmath24 , as expected .
it would now be interesting to find a physical situation where such an entropy arises from first principles , though like some initial phenomenological studies of tsallis entropy it can be also used as a parametrization scheme with @xmath28 as a parameter to fit experimental data .
the stiffness of any data may point to its preferability to tsallis - type entropies .
landsberg , `` entropies galore '' , _ braz .
j. phys . _
* 29 * , 46 ( 1999 ) a. renyi , _ probability theory _ ( north - holland , ams terdam , 1970 ) c. tsallis,_j .
phys . _ , * 52 * , 479(1988 ) p. grigolini , c. tsallis and b.j .
west,_chaos , fractals and solitons_,*13 * , 367 ( 2001 ) a.r .
plastino and a. plastino , _ j. phys .
* 27 * , 5707 ( 1994 ) c. beck , `` nonextensive statistical mechanics and particle spectra '' , hep - ph/0004225 ( 2000 ) o. sotolongo - costa et al .
, `` a nonextensive approach to dna breaking by ionizing radiation '' , cond - mat/0201289 ( 2002 ) c. wolf , `` equation of state for photons admitting tsallis statistics '' , _ fizika b _ * 11 * , 1 ( 2002 ) m. buiatti , p. grigolini and a. montagnini , _ phys .
lett . _ * 82 * , 3383 ( 1999 ) m.a .
nielsen and m. chuang , _ quantum computation and quantum information _
( cambridge u.p . ,
ny , 2000 ) a.i .
kinchin , _ mathematical foundationsof information theory _ , ( dover publications , new york , 1957 )
|
we propose a new way of defining entropy of a system , which gives a general form which may be nonextensive as tsallis entropy , but is linearly dependent on component entropies , like renyi entropy , which is extensive .
this entropy has a conceptually novel but simple origin and is mathematically easy to define by a very simple expression , though the probability distribution resulting from optimizing it gives rather complex , which is compared numerically with the other entropies .
it may , therefore , appear as the right candidate in a physical situation where the probability distribution does not suit any of the previously defined forms .
|
cosmic microwave background ( cmb ) power spectrum estimation from large datasets such as the wilkinson microwave anisotropy probe ( _ wmap _ ) @xcite presents a considerable computational challenge . with the exception of some specialized methods @xcite , obtaining the power spectrum and error bars for a generalized , large cmb dataset without introducing significant simplifications has , until now , been computationally impossible .
however , a new numerical approach based on gibbs sampling has been developed recently by jewell , levin , and anderson ( 2004 ) and wandelt , larson , and lakshminarayanan ( 2004 ) which offers the hope of overcoming these computational difficulties and allowing the analysis of large cmb datasets regardless of scanning strategy , noise characteristics and so on . in this letter
we present our results from the application of this new method to the first year _
wmap _ data .
we developed two independent , parallel codes to achieve this and a detailed description of the implementation of these codes is given in a companion publication @xcite .
the analysis results in a full , multivariate probabilistic description of the information the _ wmap _ data set contains about the power spectrum and the all - sky map of the cmb anisotropies . for the purposes of this letter we limit ourselves to a presentation of the results at each angular scale @xmath1 , leaving a full multivariate exploration of the _ wmap _ likelihood for a future publication . in 2 we give a very brief overview of our method and in 3 we present on overview of the implementation of our codes .
4 contains our results , provides some interpretation and discusses what light our method can shed on some of the anomalies in the _ wmap _ power spectrum that have been discussed in the literature ( e.g. @xcite ) .
our conclusions and future directions for this work are detailed in 5 .
in this section we present only a very brief overview of our method and codes and refer the reader to @xcite for a detailed discussion of our implementations and @xcite and @xcite for a discussion of the underlying principles .
the main difference between our approach and previous attempts at power spectrum estimation is that rather than trying to solve the maximum likelihood problem directly , we _
sample _ from the probability density of the power spectrum @xmath0 given the data @xmath4 , @xmath5 . while directly sampling from this density is difficult or impossible , we can sample from the joint posterior density @xmath6 , where @xmath7 is the cmb signal map .
expectation values of any statistic of the @xmath8 will converge to the expectation values of @xmath5 .
the theory of gibbs sampling @xcite tells us that if we can sample from the conditional densities @xmath9 and @xmath10 then iterating the following two sampling equations will , after an initial burn - in period , lead to a sample from the joint posterior @xmath11 : @xmath12 here the symbol `` @xmath13 '' denotes drawing a random realization from the density on the right .
the conditional density of the signal given the most recent @xmath8 sample is @xmath14 , where @xmath15 is the map constructed from the data @xmath4 . to sample from this density
we simply generate a gaussian variate with the required mean and covariance . the density for the @xmath8 factorizes to an inverse gamma distribution due to
the special form of @xmath16 and sampling from this is very simple . for each @xmath1
we compute @xmath17 and a ( 2@xmath1 - 1)-vector of gaussian random variates with zero mean and unit variance
. then @xmath18 . given this sample from @xmath19 , the marginalization over @xmath7 to obtain @xmath20
is trivially accomplished : just ignore the @xmath7 in the tuple @xmath21 .
of course this is not to say that the @xmath7 sample is purely ancillary in the analysis : it is also of interest to explore @xmath22 which is a full representation of the cmb signal content of the data .
the scaling of this monte carlo method is @xmath23 and corresponds to the scaling of the spherical harmonic transforms performed by the healpix algorithm .
other issues which affect our computational cost are the choice of a good preconditioner for our linear algebra solver , the correlation length between samples and the length of burn - in time required for the sampling scheme , eqs . 1 and 2 , to converge .
we perform the sampling in eq . 1 , by means of solving linear systems of equations using the conjugate gradient algorithm .
the computational cost of this method is therefore highly dependent on the choice of a good preconditioner for this system .
we experimented with several preconditioners and found examples which gave convergence in a few tens of iterations .
the number of samples we need to obtain in order to adequately describe the statistics of the power spectrum will depend upon the degree of correlation between successive samples .
a detailed discussion of these issues can be found in @xcite .
an appealing feature of the sampling approach is its ability to incorporate virtually any real - world complications , as discussed by @xcite and @xcite .
a few examples of this flexibility are applications to @xmath24 noise , asymmetric beams or arbitrary sky coverage .
foreground estimation and removal are also possible . in principle , detailed prior knowledge of all anticipated foregrounds can be included and the algorithm will then return the level of the foreground supported by the data in the map .
however , for this first analysis we limit ourselves to including two simple foreground components : ( 1 ) a stochastic model of the monopole and dipole in the maps , and ( 2 ) a stochastic model of the foreground component .
these models are encoded in terms of uniform , improper priors that express our ignorance of the monopole and dipole in the maps as well as the foreground contributions in the regions that are flagged in the foreground masks supplied by the _ wmap _ team .
traditionally this masked region is excluded from the analysis altogether . in the bayesian approach
it is modeled as a region where the foregrounds are completely unknown .
sampling from the posterior density reconstructs the signal in the masked region , based on the correlation structure discovered on the unmasked portion of the sky .
the details of these foreground models are outlined in the above papers as well as in @xcite .
we used two independently developed implementations of the algorithm to produce our results : magic @xcite and commander . these are both parallel implementations of the sampling algorithm . having results from both magic and commander allowed us to cross - check results and provided valuable redundancy .
our tests of the codes are detailed in @xcite and these showed that both codes were producing results consistent with each other . in both codes our noise model for each channel consisted of combining the published number of observations per pixel with the noise per sample for each channel and assuming the noise to be uncorrelated .
we also show in @xcite that residual correlated noise in the _ wmap _ maps has negligible effect on our results .
the data input to these codes were the first year _
maps for all 8 of the cosmologically interesting frequency bands ( 2 at q- , 2 at v- and 4 at w - band ) . for all channels we used template corrected maps following @xcite .
we performed two analyses with these data .
first we analyzed the data in the v and w channels separately by band , using unweighted averages of the channel maps and computing an effective beam window function .
for these runs we chose the conservative kp0 mask and ran single , long chains , containing 1000 samples for each band .
q - band was analyzed more aggressively , combining the channels on the likelihood level @xcite , and using the kp2 mask to assess potential foreground contamination .
these analyses will be labeled q , v and w in the following .
the second analysis uses the optimal combination of all 8 channels in the likelihood using their respective noise specifications and beam window functions
. we will refer to this joint analysis as qvw in the following .
as before , we specified a stochastic model for the monopole and dipole component as well as complete ignorance about the foregrounds within a region on the sky .
for the qvw analysis this region was defined by the more aggressive kp2 mask .
both the q and qvw analyses were done in parallel , short chains ( @xmath2510 chains , @xmath25100 samples each ) initialized with plausible starting spectra inspired by the
_ wmap _ analysis @xcite .
all of the results presented here are based on @xmath25 1000 samples for each frequency band , which required a total of @xmath2515,000 hours of cpu time figure [ fig : qvwsignal ] shows the bayes estimate ( posterior mean ) for the signal qvw analysis .
this map represents the cmb signal content in the q , v , and w bands of the _ wmap _ data .
this map shows detail where the data warrants it and smoothes where the data is poor .
the cmb signal is clearly visible outside the galaxy and the algorithm has reconstructed the signal in the galactic plane for the largest scale modes , consistent with the signal phases and correlations it has discovered on the high latitude sky .
there is insufficient information in the map for the algorithm to reconstruct the smaller scale modes in the galactic plane .
this result is a non - linear generalization of the wiener filter which does not assume prior information about the signal correlations but discovers them from the data .
figure [ fig:4ps ] shows the power spectra we obtain for q- , v- and w - band and the combined qvw analysis .
large scale features in the power spectra are consistently reproduced across all the channels . in figure [ fig:20panelcomb ]
, we show the probability distribution of the @xmath0 samples for some @xmath1 of interest .
since the original _ wmap _ analysis was released there has been some controversy regarding a lack of power at the largest angular scales ( lowest @xmath1 s ) e.g. @xcite .
it can be seen from the plots that our estimates of these @xmath8 s are consistent with the original determinations made by the _
wmap _ team , but that there are significant , non - gaussian uncertainties on our estimates which indicate that low values of @xmath0 for the largest angular scales are not so unexpected . we also over - plot the results using the _ wmap _ internal linear combination map with the kp2 mask from @xcite in our figure .
it can be seen that there is good agreement between the values obtained in our analysis and that of efstathiou . for easy reference
we show the cumulative probability distributions for the lowest @xmath1 in figure [ fig : odds ] .
one can read off the probability that the actual theory @xmath0 are lower than the prediction from any particular model .
probabilities close to 0 or 1 indicate a poor fit , or outlier .
a search over all strongly signal dominated @xmath1 from 2 to 350 using our v and w band analyses resulted in 10 outliers ( within 0.01 of 0 or 1 ) for v band ( @xmath2654 , 73 , 114 , 117 , 121 , 179 , 181 , 209 , 300 , 322 ) and 9 outliers for w band ( @xmath1=73 , 82 , 117 , 121 , 181 , 261 , 334 , 341 , 344 ) . for 349 tests
this corresponds to only a slightly higher number of outliers than expected based on counting statistics , and is entirely unremarkable if we only count those outliers that appear in both bands .
based on this @xmath1-by-@xmath1 analysis the `` bite '' in the spectrum does not seem particularly extraordinary , with only one outlier in the range of 200 - 220 .
however , all @xmath0 at @xmath27 are quite far below the best fit and will most likely be noted as a highly significant outlier in a full multivariate analysis of the joint posterior density .
we have implemented the gibbs sampling technique introduced by @xcite and @xcite , and applied it to the _ wmap _ data .
this first presentation of our results focuses on the inferred marginalized likelihoods for each @xmath1 .
we find that these are broadly consistent with previous determinations of the power spectrum . by virtue of using a bayesian analysis
we can present present a more complete picture of the uncertainties in the estimates across all relevant scales .
previous likelihood - based work to generate a full probabilistic description ( e.g. @xcite ) were limited to looking at only the lowest @xmath1 due to the computational difficulty involved .
we present an analysis that covers the region from @xmath28 to @xmath29 .
a multivariate treatment will therefore allows a rigorous assessment of the significance of anomalies in the power spectrum such as the low power on large angular scales , the `` bite '' in the power spectrum near the peak , etc .
, as well as using the full multivariate posterior density as the input to parameter estimation .
this is particularly interesting since the bayesian approach yields a _ converging _ analytic approximation to the likelihood of @xmath0 given the data , it is not necessary to construct parametric approximations to the likelihood such as the gaussian or shifted log - normal approximations .
a detailed study of these issues will be presented in a future publication .
more information can be drawn from the data in the future .
the extension to more complex foreground models , the inclusion of correlated noise and the addition of polarization to the analysis will all be of great interest , especially in view of the upcoming release of the second year of _ wmap _ data and , further in the future , the _ planck _ experiment .
we acknowledge use of the healpix software ( grski , hivon & wandelt 1998 ) and analysis package .
we also acknowledge use of the legacy archive for microwave background data analysis ( lambda ) .
h. k. e.and p. b. l. acknowledge financial support from the research council of norway , including a ph .
d. studentship for h. k.e .
this work has also received support from the research council of norway ( programme for supercomputing ) through a grant of computing time .
this work was partially performed at the jet propulsion laboratory , california institute of technology , under a contract with the national aeronautics and space administration .
this work was partially supported by an ncsa faculty fellowship for b.d.w .
this research used resources of the national energy research scientific computing center , which is supported by the office of science of the u.s .
department of energy under contract no .
de - ac03 - 76sf00098 .
bennett , c. l. _ et al . _ 2003a , , 148 , 1 bennett , c. l. _ et al . _ 2003b , , 148 ,
97 efstathiou , g. 2004 , , 348 , 885 eriksen , h. k. _ _ et al.__2004 , , accepted , astro - ph/0407028 gelfand , a. e. and smith , a. f. m. 1990 , j. am .
stat . asso .
85 , 398 grski , k. m. , hivon , e. , & wandelt , b. d. , 1999 , in evolution of large - scale structure : from recombination to garching , ed .
j. banday , r. k. sheth , & l. n. da costa ( garching , germany : european southern observatory ) , 37 hinshaw , g. _ et al . _ 2003 , , 148 , 135 jewell , j. , levin , s. , & anderson , c. h. 2004 , , 609 , 1 oh , s. p. , spergel , d. n. , & hinshaw , g. 1999 , , 510 551 slosar , a. , seljak , u. , & makarov , a. 2004 , , 69 , 123003 wandelt , b. d. , larson , d. l. , & lakshminarayanan , a. 2003 , prd accepted [ astro - ph/0310080 ] wandelt , b. d. 2004 , in `` stanford 2003 , statistical problems in particle physics , astrophysics and cosmology '' , lyons , l. _ et al . _
( eds . ) , p. 229
[ astro - ph/0401623 ]
|
we present the first results from a bayesian analysis of the _ wmap _ first year data using a gibbs sampling technique . using two independent , parallel supercomputer codes
we analyze the _ wmap _ q , v and w bands .
the analysis results in a full probabilistic description of the information the _ wmap _ data set contains about the power spectrum and the all - sky map of the cosmic microwave background anisotropies .
we present the complete probability distributions for each @xmath0 including any non - gaussianities of the power spectrum likelihood . while we find good overall agreement with the previously published _ wmap _ spectrum , our analysis uncovers discrepancies in the power spectrum estimates at low @xmath1 multipoles .
for example we claim the best - fit @xmath2cdm model is consistent with the @xmath3 inferred from our combined q+v+w analysis with a 10% probability of an even larger theoretical @xmath3 . based on our exact analysis
we can therefore attribute the `` low quadrupole issue '' to a statistical fluctuation .
|
glycine ( nh@xmath0ch@xmath0cooh ) , the simplest amino acid , has been identified in cometary @xcite and meteoritic samples @xcite , however it is not yet known whether the species is formed in these solid bodies , in the interstellar medium ( ism ) , or both . indeed , a detection in the ism has so far proven ellusive @xcite .
although chemical models predict the formation of glycine in the ism , in low abundance ( see , e.g. @xcite and refs . therein ) , it has been difficult to observationally constrain them without a detection of the species .
the relative likelihood of different possible formation routes can , however , be constrained by careful observation of the reactants used in the chemical models to form glycine .
one pathway which has garnered significant interest is the formation of glycine through the reaction of hydroxylamine ( nh@xmath0oh ) , or its protonated and ionized derivatives , with acetic acid ( ch@xmath3cooh ) , a known interstellar molecule @xcite .
ionization or protonation of nh@xmath0oh under interstellar conditions should be efficient @xcite , and subsequent laboratory work has demonstrated the formation of glycine from nh@xmath0oh and its ionized and protonated forms @xcite . although recent theoretical work has suggested that these gas - phase routes forming glycine through these reactions are inefficient under interstellar conditions @xcite , condensed - phase surface - mediated reactions remain a possibility , and observational constraints of the precursors are still desirable for the refinement of models .
while acetic acid is readily - constrained , nh@xmath0oh has yet to be observed in the ism . in recent years , laboratory work has shown that a number of pathways exist which result in efficient formation of nh@xmath0oh on grain surfaces .
@xcite demonstrated the formation of nh@xmath0oh in h@xmath0o - nh@xmath3 ices after uv - irradiation .
formation by successive hydrogenation of no has been shown to proceed efficiently , barrierlessly , and in high yield @xcite .
most recently , @xcite demonstrated the efficient formation of nh@xmath0oh via oxidation of nh@xmath3 in ices with a low barrier to activation ( of order 1000 k ) .
earlier modeling work by @xcite also suggested that nh@xmath0oh is formed in high abundance in interstellar ices , and is subsequently liberated into the gas - phase during the warm - up period of emerging hot cores and hot corinos in readily - detectable quantities .
observational efforts by @xcite , however , failed to detect nh@xmath0oh toward a selection of such sources known to be rich in complex organic material ( e.g. sgr b2(n ) and orion - kl ) establishing upper limits of n@xmath9 .
refinement of the @xcite model in subsequent work @xcite , as well as limiting modeling presented in the aforementioned laboratory studies , resulted in predicted gas - phase abundances in line with the upper limits established by @xcite . yet , with the inclusion of the h + hno @xmath10 hnoh pathway into the model , as described by the laboratory work of @xcite , condensed phase abundances of nh@xmath0oh again approach n@xmath11 , with gas - phase abundances of n@xmath12 .
the @xcite model largely assumes that the release of nh@xmath0oh into the gas - phase is a gradual process dominated by the warm - up of the hot core .
both the laboratory work and the models , however , predict nh@xmath0oh is initially formed in large abundance at very cold ( @xmath13 k ) temperatures and early in the evolution of these sources
. thus , the most optimistic source for a detection of gas - phase nh@xmath0oh is one where the reservoir of condensed - phase nh@xmath0oh formed at low temperatures is liberated _ en masse _ into the gas - phase prior to release by thermal mechanisms .
shocked regions displaying high degrees of molecular complexity likely represent this best - case scenario . in these regions ,
complex molecules are formed efficiently in ices at low temperatures , but are not otherwise liberated into the gas - phase except by thermal desorption at much greater temperatures . when these ices are subjected to shocks ,
however , the mantle is non - thermally ejected into the gas - phase , resulting in large abundances of relatively cool ( @xmath14 k ) , complex organic material ( see , e.g. , @xcite ) .
one of the most prominent of these regions is the young protostellar outflow l1157 .
numerous recent studies report high degrees of molecular complexity arising from shocked regions within the outflow , which originates in cold , quiescent gas around the protostar ( see , e.g. , @xcite and @xcite ) . here , we present deep searches for nh@xmath0oh using the caltech submillimeter observatory ( cso ) at @xmath15 mm and the combined array for research in millimeter - wave astronomy ( carma ) at @xmath16 mm toward the l1157 outflow .
we report non - detections of nh@xmath0oh in both searches . in order to derive upper limits to the column density of nh@xmath0oh
, we use transitions of ch@xmath3oh observed with the cso to constrain the kinetic temperature and density in the shocked gas using a radiative - transfer approach , and carma images of ch@xmath3oh to determine the size of the shocked gas .
finally , we estimate upper limits to the abundance of nh@xmath0oh and discuss possible implications .
l1157 is a dark cloud in cepheus located a distance of @xmath17250 pc @xcite , and contains a prototypical shocked bipolar outflow from a class 0 protostar .
it has been the subject of great interest in the last twenty years , with numerous studies investigating the physical conditions within the source . due to the variety of methods used for these studies , direct comparisons between results
is challenging .
nevertheless , an overall picture does come into view , and this general description will be sufficient for the discussion presented here . originating in the cold , quiescent gas ( @xmath18 k , @xcite ) surrounding the class 0 protostar l1157-mm ,
the southern lobe of the accelerated outflow ( @xmath19 k , @xcite ) has undergone two major shocks , referred to as l1157-b1 and l1157-b2 .
the b1 shock is younger and warmer ( @xmath172000 yr , @xmath20 k ) than the b2 shock ( @xmath174000 yr , @xmath21 k ) , and many complex chemical species are observed in enhanced abundance toward both shocks due to non - thermal desorption from grains @xcite . while the absolute values of these physical parameters vary somewhat within the literature , this qualitative picture and enhancement in chemical abundance is consistently reported .
the spectrum toward l1157 obtained with the cso was collected over 8 nights in 2012 july , august , and september , and 7 nights in 2014 september and december .
the telescope was pointed at the b1 shocked region at @xmath22(j2000 ) = 20@xmath2339@[email protected] , @xmath26(j2000 ) = 68@xmath2701@[email protected] and the b2 shocked region at @xmath22(j2000 ) = 20@xmath2339@xmath2413@xmath25 , @xmath26(j2000 ) = 68@xmath2700@xmath2837@xmath29 ( see figure [ methanolcsobeam ] ) and spectra were adjusted to a @xmath30 = 1.75 km s@xmath31 .
a small subset of scans toward b1 were obtained at a second position offset by -4 in declination .
the cso 230/460 ghz sidecab double - sideband ( dsb ) heterodyne receiver was used in fair weather ( @xmath32 0.1 0.25 ) which resulted in typical system temperatures of 300
500 k. the backend was a fast fourier - transform spectrometer which provided 1 ghz of dsb data at 122 khz resolution ( @xmath170.2 km s@xmath31 at 230 ghz ) .
the total frequency coverage was 188 193 ghz , 200 205 ghz , 237 243 ghz , and 249 255 ghz toward b1 , and 237 243 ghz and 249 255 ghz toward b2 .
the targeted transitions of nh@xmath0oh are given in table [ nh2ohparams ] .
a chopping secondary mirror with a throw of 4 was used for on - off calibration .
pointing was performed every @xmath172 hours and typically converged to within 1 2 .
the raw data were calibrated using the standard chopper wheel calibration method resulting in intensities on the atmosphere - corrected @xmath33 temperature scale .
these were then corrected to the main beam temperature scale , @xmath34 , where @xmath35 . for these observations ,
the main beam efficiency was @xmath36 .
the spectra were collected in dsb mode at a variety of if settings to allow for a robust deconvolution .
the class package from the gildas suite of programs was used for the data reduction .
spurious signals were removed from the spectra , which were then baseline subtracted using a polynomial fit . the standard class deconvolution routine was used to generate single - sideband data .
the spectra were then hanning smoothed to a resolution of @xmath171.4 km s@xmath31 .
the average full - width half - maximum linewidth in b1 was @xmath177.9 km s@xmath31 , and in b2 was @xmath175.2 km s@xmath31 .
l c c c c c c 2@xmath37 1@xmath38 & 100683.58(20 ) & 15.204 & 0.520 & @xmath39 & -5.9079 & @xmath40 + 2@xmath41 1@xmath42 & 100748.23(20 ) & 7.2527 & 0.694 & @xmath43 & -5.7821 & @xmath40 + 2@xmath38 1@xmath44 & 100807.62(20 ) & 15.213 & 0.520 & @xmath39 & -5.9062 & @xmath40 + & + 5@xmath45 4@xmath46 & 251677.3666(78 ) & 44.194 & 1.665 & @xmath47 & -4.5515 & @xmath48 + 5@xmath49 4@xmath50 & 251734.8061(82 ) & 163.51 & 0.625 & @xmath51 & -4.9771 & @xmath52 + 5@xmath53 4@xmath49 & 251734.8061(82 ) & 163.51 & 0.625 & @xmath51 & -4.9771 & @xmath52 + 5@xmath54 4@xmath55 & 251780.3167(70 ) & 107.85 & 1.110 & @xmath56 & -4.7270 & @xmath57 + 5@xmath55 4@xmath58 & 251780.3178(70 ) & 107.85 & 1.110 & @xmath56 & -4.7270 & @xmath57 + 5@xmath59 4@xmath60 & 251811.9913(70 ) & 68.081 & 1.457 & @xmath61 & -4.6088 & @xmath62 + 5@xmath60 4@xmath63 & 251813.8609(70 ) & 68.081 & 1.457 & @xmath61 & -4.6087 & @xmath62 + 5@xmath64 4@xmath65 & 251838.4937(79 ) & 36.261 & 1.735 & @xmath66 & -4.5329 & @xmath67 + 5@xmath46 4@xmath68 & 251987.1350(78 ) & 44.239 & 1.665 & @xmath47 & -4.5498 & @xmath57 + [ nh2ohparams ]
a total of 89.3 hours of observations were conducted with the carma 15-element array in c - configuration ( 2013 may ) , d - configuration ( 2012 october , november ) , and e - configuration ( 2012 august ) at @xmath2 mm .
the phase center for these observations was @xmath22(j2000 ) = 20@xmath2339@[email protected] , @xmath26(j2000 ) = 68@xmath2701@[email protected] .
the carma correlator was used in its 62 mhz bandwidth , 3-bit mode providing 255 channels across the band for a resolution of 243 khz or 0.7 km s@xmath31 .
the three targeted transitions of nh@xmath0oh are given in table [ nh2ohparams ] .
mwc349 and neptune were used as primary flux calibrators ; the passband calibrators were 1635 + 381 , 2232 + 117 , 0102 + 584 , 1743 - 038 , 2015 + 372 , and 3c84 .
the gain calibrator was 1927 + 739 .
data reduction was completed using the miriad package .
data were flagged for antennas that were offline or malfunctioning during observation or when the phases in the calibrators had deviations greater than thirty degrees from a smooth trend line .
image cleaning was also performed using miriad .
the robust factor tended towards natural weighting with cell size set to 0.4 and image size of 1024 pixels .
no channel averaging was applied for the maps shown ; however , channel averaging to 1.4 km s@xmath31 was used on windows targeting hydroxylamine transitions .
clean regions were drawn with the polygon tool around clear emission and cleaned to an average noise of 0.7 mjy beam@xmath31 .
the restoring beam was typically @xmath173.4 @xmath69 3.2 .
the full set of observations toward l1157 carma included a host of other molecular species , and will be presented in a follow - up paper by dollhopf et al .
an accurate determination of nh@xmath0oh column density upper limits in the b1 and b2 shocks requires the kinetic temperature and density in these regions . to obtain these
, we fit the observed ch@xmath3oh emission in the same spectral window ( table [ ch3ohparams ] ) to a non - lte model of the outflow and shock , based on a physical model derived from our our carma observations , using the radex code @xcite .
l c c c c c c + & + 2@xmath41 1@xmath42 + + & 96741.375(5 ) & 6.9650 & 1.617 & @xmath43 & -5.4675 & @xmath70 + & + 5@xmath45 4@xmath46 + + & 239746.253 & 49.059 & 3.885 & @xmath71 & -4.2468 & @xmath72 + 5@xmath64 4@xmath65 + + & 241791.431 & 34.817 & 4.043 & @xmath66 & -4.2184 & @xmath73 + 5@xmath54 4@xmath55 + + & 241832.91(20 ) & 84.618 & 2.578 & @xmath56 & -4.4137 & @xmath74 + 5@xmath55 4@xmath58 - - & 241832.91(20 ) & 84.618 & 2.578 & @xmath75 & -4.4137 & @xmath76 + 5@xmath59 4@xmath60 - - & 241842.324 & 72.530 & 3.415 & @xmath77 & -4.2915 & @xmath78 + 5@xmath60 4@xmath63 + + & 241887.704 & 72.533 & 3.415 & @xmath79 & -4.2912 & @xmath80 + 5@xmath46 4@xmath68 - - & 243915.826 & 49.661 & 3.885 & @xmath81 & -4.2243 & @xmath82 + 11@xmath83 10@xmath84 + + & 250506.980 & 153.10 & 10.63 & @xmath85 & -4.0728 & @xmath86 + 8@xmath87 8@xmath88 - + & 251517.262 & 133.36 & 7.308 & @xmath89 & -4.0990 & @xmath62 + 7@xmath90 7@xmath91 - + & 251641.667 & 114.79 & 6.279 & @xmath92 & -4.1099 & @xmath93 + 6@xmath54 6@xmath59 - + & 251738.520 & 98.546 & 5.224 & @xmath94 & -4.1272 & @xmath72 + 5@xmath55 5@xmath60 - + & 251811.882 & 84.618 & 4.126 & @xmath95 & -4.1567 & @xmath96 + 4@xmath58 4@xmath63 - + & 251866.579 & 73.012 & 2.954 & @xmath97 & -4.2144 & @xmath98 + 5@xmath54 5@xmath59
+ - & 251890.901 & 84.619 & 4.125 & @xmath99 & -4.1564 & @xmath100 + 6@xmath90 6@xmath91 + - & 251895.728 & 98.545 & 5.220 & @xmath101 & -4.1266 & @xmath102 + 4@xmath55 4@xmath60 + - & 251900.495 & 73.012 & 2.953 & @xmath103 & -4.2143 & @xmath57 + 3@xmath104 3@xmath105 - + & 251905.812 & 63.727 & 1.641 & @xmath106 & -4.3603 & @xmath57 + 3@xmath58 3@xmath63
+ - & 251917.042 & 63.727 & 1.641 & @xmath107 & -4.3603 & @xmath102 + 7@xmath87 7@xmath88 + - & 251923.631 & 114.79 & 6.272 & @xmath108 & -4.1089 & @xmath109 + 8@xmath110 8@xmath111 + - & 251984.702 & 133.36 & 7.293 & @xmath112 & -4.0975 & @xmath96 + & + + & + 2@xmath113 1@xmath113 & 96739.362(5 ) & 12.541 & 1.213 & @xmath114 & -5.5922 & @xmath115 + 2@xmath41 1@xmath41 & 96744.550(5 ) & 20.090 & 1.617 & @xmath116 & -5.4675 & @xmath52 + 2@xmath38 1@xmath44 & 96755.511(5 ) & 28.011 & 1.244 & @xmath117 & -5.5811 & @xmath118 + & + 5@xmath64 4@xmath65 & 241700.219 & 47.934 & 4.040 & @xmath119 & -4.2192 & @xmath74 + 5@xmath120 4@xmath121 & 241767.224 & 40.391 & 3.882 & @xmath122 & -4.2362 & @xmath123 + 5@xmath55 4@xmath58 & 241843.646 & 82.531 & 2.587 & @xmath124 & -4.4120 & @xmath102 + 5@xmath46 4@xmath68 & 241879.073 & 55.871 & 3.980 & @xmath125 & -4.2248 & @xmath115 + 5@xmath126 4@xmath127 & 241904.152 & 60.725 & 3.399 & @xmath128 & -4.2932 & @xmath129 + 5@xmath60 4@xmath63 & 241904.645 & 57.069 & 3.356 & @xmath130 & -4.2987 & @xmath115 + 2@xmath41 1@xmath131 & 254015.340 & 20.090 & 0.499 & @xmath132 & -4.7208 & @xmath133 + [ ch3ohparams ] first , we determined the approximate spatial extent and location of the shocked gas from the ch@xmath3oh emission observed with carma .
figure [ methanolcsobeam ] shows the cso beam at the targeted positions overlaid on our carma observations of ch@xmath3oh and hnco at @xmath2 mm .
the pointing position for b1 was chosen from the literature prior to our carma observations , and is roughly centered on the shock front , rather than the peak of the shock emission .
the pointing position for b2 was chosen after we had acquired the carma observations , and is centered on the peak of the shocked hnco emission . in both cases ,
the shock is nearly completely contained within the beam , and we estimate a reasonable equivalent size of @xmath1712 for the shock , and @xmath1720 for the second component , within our @xmath1730 cso beam . in b1 , the shock is not centered in the beam , and thus any emission will suffer more significant fall - off effects than in b2 .
we made the initial assumption that the two components in our fit represented the shocked gas and the outflow . under these assumptions , and based on our carma observation as well as prior studies @xcite , we constrained the fit of the ch@xmath3oh emission with the following assumptions : 1 .
the kinetic temperature of the shock was no less than that of the outflow .
the density of the shock was no less than that of the outflow .
the column density of ch@xmath3oh was no less in the shock than in the outflow .
assumptions ( 1 ) and ( 2 ) are well - substantiated in the literature , both toward l1157 @xcite and in shocked environments in general @xcite , and assumption ( 3 ) agrees with the previous findings of @xcite . using these constraints , and the two - source component model described earlier , we performed a reduced-@xmath134 analysis of a grid of models for the cso observations , simultaneously fitting @xmath135- and @xmath136-ch@xmath3oh following the methods described in @xcite .
a thorough discussion of the uncertainties in the fitting method is provided in appendix [ appa ] .
based on these uncertainties , and the assumptions described above , we find the parameters in table [ methanolradex ] represent a best - fit to the data , with reduced-@xmath134 values of 2.50 and 1.77 for b1 and b2 , respectively .
these values are likely due to the simplicity of our model relative to the complexity of the source .
literature sources ( see , e.g. , @xcite ) suggest a structure with three or more components is more realistic , but such additional components would not be well - constrained in our case due to the limited number of observed ch@xmath3oh transitions
. additionally , there are likely quite large temperature and density gradients , especially in the newer , more compact b1 shock , which will contribute additional error to the fit which is not accounted for in our analysis . finally , ch@xmath3oh excitation is sensitive to the far - ir radiation field present , which we have assumed as standard galactic background , any deviation from which would further impact the accuracy of the model and increased the ultimate @xmath134 .
simulated non - lte spectra of ch@xmath3oh from these results toward b1 and b2 are provided in figures [ methanolresults ] and [ methanolresultsb2 ] , respectively , overlaid on observations and corrected for beam efficiency and a 12 source size .
the ch@xmath3oh column densities and h@xmath0 densities found here are in relatively good agreement with previous observations of the source @xcite .
the derived values of @xmath137 for the warmer component , which we ascribe to the shocks , agrees well with previous measurements in b2 @xcite , while the upper range of our derived value in b1 falls at the lower edge of previous measurements .
the kinetic temperature of the second component ( @xmath138 k ) is significantly lower than previous measurements of both the shock and the outflow ( @xmath139 k ) , indicating that it is likely not probing the outflow as we had originally presumed . instead , toward b1 ,
this component is more likely related to the @xmath140 component described by @xcite as remnants from the gas in which the earlier b2 shock was formed , also offering a possible explanation for the component s presence toward b2 .
the @xmath140 component has been reported to be both cold ( @xmath141 k ) and extended @xcite , in qualitatively good agreement with our fit .
c c c c c c @xmath142 ( @xmath143 ) & @xmath144 & @xmath144 & & @xmath145 & @xmath145 + @xmath137 ( k ) & 60(20 ) & 10(3 ) & & 50(15 ) & 10(3 ) + @xmath146 ( @xmath5 ) & @xmath147 & @xmath148 & & @xmath147 & @xmath149 + [ methanolradex ]
despite deep searches , we find no conclusive evidence for nh@xmath0oh emission at @xmath2 mm or @xmath1 mm in either the carma or the cso data , respectively . to calculate appropriate upper limits for nh@xmath0oh in the both datasets
, we assume that nh@xmath0oh , if present in the gas phase , originates in the warm , shocked gas traced by the warm , dense ch@xmath3oh component .
the rationale for this assumption is discussed in
[ discussion ] . near the temperatures derived from our radex fit in the warm component toward b1 and b2 ,
the strongest nh@xmath0oh transition is the 5@xmath64
4@xmath65 at 251838 mhz .
as noted in table [ nh2ohparams ] , the critical density ( @xmath150 ) for this transition is @xmath151 @xmath143 .
this value is based on the assumption that the collisional coefficients for nh@xmath0oh are similar to those of @xmath135-ch@xmath3oh . while not exact , the similar mass , dipole moments , molecular size , and energy level structures make this a reasonable approximation within the context of the following discussion . given the densities derived from our radex fit ( 3 6 @xmath152 @xmath143 ) , we therefore assume that , at least for this transition , lte is a reasonable approximation for determining nh@xmath0oh upper limits and thus @xmath153 .
under these conditions , we derive 1@xmath154 upper limit column densities for nh@xmath0oh of @xmath155 @xmath5 and @xmath156 @xmath5 in b1 and b2 , respectively .
@xcite derive a co column density of @xmath157 @xmath5 in the @xmath158 component of the b1 shock , arising from the shocked gas and covering the entire region , assuming a 20 source size . assuming the co is homogeneously distributed over the region
, this gives a source - averaged co column of @xmath159 @xmath5 for the 12 source size used here .
taking @xmath160 ( c.f .
@xcite and @xcite ) , this results in an h@xmath0 column density of @xmath17@xmath161 @xmath5 .
a similar argument using the co column density of @xmath162 @xmath5 for b2 found by @xcite also results in an h@xmath0 column density of @xmath17@xmath161 @xmath5 .
this gives upper limits to the fractional abundances of nh@xmath0oh of @xmath163 and @xmath164 for b1 and b2 , respectively . to calculate appropriate upper limits for nh@xmath0oh in the carma data
, we again assume that nh@xmath0oh , if present in the gas phase , originates in the warm , shocked gas traced by the warm , dense ch@xmath3oh component .
the spectra extracted from a 12 beam centered on these locations are shown in figure [ nh2ohcarma ] . while there is an unidentified feature at the frequency of the @xmath165 nh@xmath0oh transition , if this were truly nh@xmath0oh emission , the other transitions in the observed window would be of equal or greater intensity .
we measure the rms to be 1.6 mk and 2.9 mk in the b1 and b2 spectra . at the derived temperatures , the @xmath166 transitions at 100748 mhz
is predicted to be strongest .
as both b1 and b2 have densities higher than @xmath150 for this transition ( see table [ nh2ohparams ] ) , we again derive an upper limit to the column density assuming lte conditions . following the same procedure as for the cso data ,
we find a 1@xmath154 upper limit of nh@xmath0oh of @xmath167 @xmath5 for b1 and @xmath168 @xmath5 for b2 .
this gives upper limits to the fractional abundances of nh@xmath0oh of @xmath169 and @xmath8 for b1 and b2 , respectively .
all results are summarized in table [ nh2ohcdresults ] .
c c c c c c @xmath170 ( @xmath5 ) & @xmath155 & @xmath156 & & @xmath167 & @xmath168 + @xmath171 & @xmath172 & @xmath164 & & @xmath173 & @xmath8 + [ nh2ohcdresults ]
in [ results ] , we assumed that nh@xmath0oh , if present , would arise predominantly the warm , shocked regions b1 or b2 , rather than the molecular outflow .
@xcite show that nh@xmath0oh is thermally liberated from their laboratory samples between 160 180 k , whereas for realistic interstellar ices and conditions , temperatures above @xmath17110 k are likely sufficient @xcite .
both our non - lte fits , and the literature values discussed earlier have shown that temperatures in the targeted regions are below these thresholds , and thus a thermal mechanism for the desorption of nh@xmath0oh from grain surfaces in l1157 is unlikely . thus , non - thermal desorption in the shocks should be the dominant mechanism for nh@xmath0oh liberation , and any nh@xmath0oh emission should trace these shocked regions .
the upper limit column densities established with the carma data are equivalent to those established by @xcite of @xmath174 @xmath5 .
the limits established by @xcite , however , are beam - averaged column densities , whereas the carma measurements presented here provide a more well - constrained source size , and thus a more robust upper limit . nevertheless , the upper limits to relative abundance found here are significantly higher than those determined by @xcite of @xmath1710@xmath175 due to the low h@xmath0 column in the region .
these low columns are common in other chemically - complex shocked regions as well , such as a host of galactic center clouds studied by @xcite ( @xmath176 @xmath5 ) , and other shocked outflows like bhr 71 ( @xmath177 @xmath5 ; @xcite ) . yet , these regions likely represent the best - case scenario for a gas - phase detection of nh@xmath0oh .
such a detection is critical to accurately constrain the application of laboratory results , which show nh@xmath0oh is a significant player in grain - surface nitrogen chemistry , to chemical models .
the deep search presented here stretched the capabilities of the cso and carma to the limit ; searches deep enough to be sensitive to the relative abundances predicted by laboratory work and modeling are simply time - prohibitive on these types of instruments .
the @xcite model , without the addition of the h + hno @xmath10 hnoh pathway , predicts peak condensed - phase nh@xmath0oh abundances of @xmath17@xmath178 , which , if liberated _ en masse _ in a shock , is a detectable population , but likely only with the sensitivity and spatial resolution of alma .
the upper limits established here , however , are several orders of magnitude lower than both the condensed - phase and gas - phase populations predicted by the augmented model which includes this hydrogenation pathway .
thus , our results demonstrate that further efforts are needed both in modeling and in the laboratory , to identify and fully - constrain both the formation and the destruction pathways for nh@xmath0oh .
bam gratefully acknowledges funding by an nsf graduate research fellowship during initial portions of this work .
the authors thank the anonymous referee for comments which increased the quality of this work .
we thank s. radford for support of the cso observations .
the national radio astronomy observatory is a facility of the national science foundation operated under cooperative agreement by associated universities , inc .
portions of this material is based upon work at the caltech submillimeter observatory , which was operated by the california institute of technology under cooperative agreement with the national science foundation ( ast-0838261 ) .
support for carma construction was derived from the gordon and betty moore foundation , the kenneth t. and eileen l. norris foundation , the james s. mcdonnell foundation , the associates of the california institute of technology , the university of chicago , the states of california , illinois , and maryland , and the national science foundation .
the uncertainty in the observations is given as equation [ uncertainty ] , modified from equation a1 in @xcite .
@xmath183 is the uncertainty in the beam - filling factor .
the size of the telescope beam varies by @xmath174 across the band , is well - described , and is calculated independently at each transition frequency , thus we assume no contribution to @xmath183 from the telescope beam . however , the sizes of the two source components are not a varied parameter , and are only modestly well - described by our carma observations . further ,
in the case of b1 , the shock is positioned off - center of the beam , making the emission more sensitive to fall - off effects that are not accounted for explicitly in the calculations .
we therefore assume a factor of two uncertainty in source size to take both issues into consideration , which is reflected in @xmath183 .
these result in net uncertainties of @xmath1732% in the absolute intensity of the observed transitions , with the largest contribution from the absolute flux calibration .
@xmath134 values outside of the extremes represented by these uncertainties more than double those of the best - fit parameters , so we take @xmath1732% to be a reasonable estimate of the overall uncertainty in n@xmath184 , @xmath146 , and @xmath137 .
angelelli , f. , aschi , m. , cacace , f. , pepi , f. , & de petris , g. 1995 , j. phys .
, 99 , 6551 .
arce , h.g .
, santiago - garca , j. , jrgensen , j.k , tafalla , m. , & bachiller , r. 2008 , apjl , 681 , l21 .
bachiller , r. , martn - pintado , j. , & fuente , a. 1993 , apjl , 417 , l45 .
bachiller , r. , liechti , s. , walmsley , c.m .
, & colomer , f. , 1995 , a&a , 295 , l51 .
bachiller , r. & prez gutieerrez , m. 1997 , apj , 487 , l93 barrientos , c. , redondo , p. , largo , l. , rayon , v.m . ,
largo , a. 2012 , apj , 748 , 99 .
blagojevic , v. , petrie , s. , & bohme , d.k .
2003 , mnras , 339 , l7 .
bolatto , a.d . , wolfire , m. , & leroy , a.k .
2013 , ara&a , 51 , 207 .
boulet , p. , gilardoni , f. , weber , j. , chermette , h. , & ellinger , e. 1999 , chem .
phys . , 244 , 163 .
codella , c. , fontani , f. , ceccarelli , c. , et al .
2015 , mnras , 449 , l11 .
collings , m.p . , anderson , m.a . ,
chen , r. , et al .
2004 , mnras , 354 , 1133 .
congiu , e. , fedoseev , g. , ioppolo , s. , et al .
2012a , apjl , 750 , l12 .
congiu , chaabouni , h. , laffon , c. , et al . 2012b , j. chem .
phys . , 137 , 054713 .
crockett , n.r . ,
bergin , e.a , neill , j.l .
, et al . 2014 , apj , 787 , 112 .
cunningham , m.r . ,
jones , p.a . , godfrey , p.d .
, et al . 2007 , mnras , 376 , 1201 .
elsila , j.e . , glavin , d.p . , & dworkin , j.p .
2009 , meteor . & planet .
, 44 , 1323 .
fedoseev , g. , ioppolo , s. , lamberts , t. , et al .
2012 , j. chem .
phys . , 137 , 054714 .
garay , g. , khnenkamp , i. , bourke , t.l .
, rodrguez , l.f . , & lehtinen , k.k .
1998 , apj , 509 , 768 .
garrod , r.t . ,
widicus weaver , s.l .
, & herbst , e. 2008 , apj , 682 , 283 .
garrod , r.t .
2013 , apj , 765 , 60 .
gueth , f. , guilloteau , s. , & bachiller , r. 1996 , a&a , 307 , 891 .
he , j. , vidali , g. , lemaire , j .- l . , &
garrod , r.t .
2015 , apj , 799 , 49 .
ioppolo , s. , fedoseev , g. , minissale , m. , congiu , e. , dulieu , f. , & linnartz , h. 2014 , pccp , 16 , 8270 .
jones , p.a .
, cunningham , m.r . , godfrey , p.d .
, & cragg , d.m .
2007 , mnras , 374 , 579 .
kuan , y .- j . ,
charnley , s.b . ,
huang , h .- c . ,
tseng , w .- l . , & kisiel , z. 2003 , apj , 593 , 848 .
kvenvolden , k , lawless , j. , pering , k. , et al . 1970 , nature , 228 , 923 .
largo , l. , rayn , v.m . ,
barrientos , c. , largo , a. , & redondo , p. 2009 , chem .
lett . , 476 , 174 .
looney , l.w . ,
tobin , j.j .
, kwon , w. 2007 , apjl , 670 , l131 .
lefloch , b. , cabrit , s. , busquet , g. , et al .
2012 , apj , 757 , l25 .
mehringer , d.m . ,
snyder , l.e . , miao , y. , & lovas , f.j .
1997 , apjl , 480 , l71 .
mendoza , e. , lofloch , b. , lpez - sepulcre , a. , et al .
2014 , mnras , 445 , 151 .
minissale , m. , fedoseev , g. , congiu , e. , ioppolo , s. , dulieu , f. , & linnartz , h. 2014 , pccp , 16 , 8257 .
morino , i. , yamada , k.m.t . ,
klein , h. , et al .
2000 , j. mol .
struct . , 517 , 367 .
mller , h.s.p . ,
menten , k. , & mder , h. 2004 , a&a , 428 , 1019 .
mller , h.s.p . ,
schlder , f. , stutzki , j , & winnewisser , g. 2005 , 742 , 215 .
pulliam , r.l . ,
mcguire , b.a . , &
remijan , a.j .
2012 , apj , 751 , 1 .
rabli , d. & flower , d.r .
2010 , mnras , 406 , 95 .
requena - torres , m.a .
, martn - pintado , j. , rodrguez - franco , a. , martn , s. , rodrguez - fernndez , n.j .
, & de vicente , p. 2006 , a&a , 455 , 971 .
schier , f.l . , van der tak , f.f.s . , van dishoeck , e.f . ,
& black , j.h .
2005 , a&a , 432 , 369 .
snow , j.l . ,
orlova , g. , blagojevic , v. , & bohme , d.k .
2007 , jacs , 129 , 9910 .
snyder , l.e .
, lovas , f.j . ,
hollis , j.m . ,
2005 , apj , 619 , 914 .
sugimura , m. , takahiro , y. , takeshi , s. , et al .
2010 , pasj , 63 , 459 .
tafalla , m. & bachiller , r. 1995 , apjl , 443 , l37 .
tsunekawa , j. 1972 , j. phys .
jpn . , 33 , 167 .
xu , l .- h . & lovas , f.j .
1997 , j. phys .
chem . ref .
, 26 , 17 .
van der tak , f.f.s . , black , j.h .
, schier , f.l . ,
jansen , d.j .
, & van dischoeck , e.f .
2007 , a&a , 468 , 627 .
van dishoeck , e.f . & blake , g.a .
1998 , ara&a , 36 , 317 .
zheng , w. & kaiser , r.i .
2010 , j. phys .
a , 114 , 5251 .
|
a deep search for the potential glycine precursor hydroxylamine ( nh@xmath0oh ) using the caltech submillimeter observatory ( cso ) at @xmath1 mm and the combined array for research in millimeter - wave astronomy ( carma ) at @xmath2 mm is presented toward the molecular outflow l1157 , targeting the b1 and b2 shocked regions .
we report non - detections of nh@xmath0oh in both sources .
we a perform non - lte analysis of ch@xmath3oh observed in our cso spectra to derive kinetic temperatures and densities in the shocked regions . using these parameters , we derive upper limit column densities of nh@xmath0oh of @xmath4 @xmath5 and @xmath6 @xmath5 toward the b1 and b2 shocks , respectively , and upper limit relative abundances of @xmath7 and @xmath8 , respectively .
|
the production of an electron - positron pair by a photon in an atomic field is one of the most important processes of qed . because of its importance for various applications ,
see refs .
@xcite , this process has been investigated in numerous theoretical and experimental papers .
the cross section of the process in the born approximation is known for arbitrary energy @xmath4 of the incoming photon , refs .
@xcite ( we set @xmath5 throughout the paper ) . for heavy atoms ,
it is necessary to take into account the higher - order terms of the perturbation theory with respect to the parameter @xmath6 ( coulomb corrections ) , where @xmath7 is the atomic charge number , @xmath8 is the fine - structure constant , and @xmath9 is the electron charge . the formal expression for the coulomb corrections , exact in @xmath6 and @xmath4 , was derived in ref .
however , the numerical computations based on this expression becomes more and more difficult when @xmath4 is increasing , and the numerical results have been obtained so far only for @xmath10 mev , ref .
@xcite .
fortunately , in the high - energy region @xmath11 ( @xmath1 is the electron mass ) , a completely different approach , which greatly simplifies the calculations , can be used . as a result of this approach ,
a simple expression for the coulomb corrections was obtained in refs.@xcite in the leading approximation with respect to @xmath3 .
however , this result has good accuracy only at energies @xmath12mev . for a long time
, the description of the coulomb corrections for the total cross section at intermediate photon energies ( @xmath13 mev ) was based on the expression obtained in ref .
this expression is actually an extrapolation of the results obtained at @xmath14mev .
recently , the first corrections of the order of @xmath3 to the spectrum , as well as to the total cross section , of @xmath15 photoproduction in a strong atomic field were derived in ref .
the correction to the spectrum was obtained in the region where both produced particles are relativistic .
it turns out that this correction is antisymmetric with respect to replacement @xmath16 , where @xmath17 and @xmath18 are the energy of the positron and the electron , respectively , so that the correction to the total cross section comes from the region close to the end of spectrum where @xmath19 or @xmath20 . in ref .
@xcite , the correction to the total cross section was obtained by means of the dispersion relation for the forward delbrck scattering amplitude .
the account for this correction leads to agreement between the theoretical prediction and available experimental data at intermediate photon energies , ref .
the electron ( positron ) spectrum in the process of @xmath15 photoproduction in a strong coulomb field in the case @xmath19 or @xmath20 and @xmath11 was investigated in ref .
it was shown that the coulomb corrections drastically differ from those obtained in the region where @xmath21 and @xmath22 .
integration of the spectrum in ref .
@xcite has confirmed the result for the correction to the total cross section obtained in ref .
@xcite by means of the dispersion relation . in the present paper , we calculate , exactly in @xmath23 , the next - to - leading correction with respect to @xmath3 to the differential cross section of electron - positron pair production by a high - energy photon in a strong atomic field .
the correction , being the odd function of @xmath24 , gives rise to the charge asymmetry in this process .
the leading term of the cross section obtained in refs .
@xcite is the even function of @xmath6 and does not contribute to the charge asymmetry .
for the photon energy below the threshold of @xmath25-meson photoproduction , we also estimate the contribution of the compton - type amplitude to the @xmath15 photoproduction cross section .
the corresponding term contributes to the charge asymmetry as well .
the cross section of @xmath15 pair production by a photon in an external field reads ( see , e.g. , ref .
@xcite ) @xmath26 where @xmath27 , @xmath28 and @xmath29 are the electron and positron momenta , respectively .
the matrix element @xmath30 , corresponding to the diagram shown in fig.[fig : dia1 ] has the form @xmath31 here @xmath32 is a positive - energy solution and @xmath33 is a negative - energy solution of the dirac equation in the external field , @xmath34 and @xmath35 enumerate the independent solutions of the dirac equation , @xmath36 is the photon polarization vector , @xmath37 is the photon momentum , @xmath38 are the dirac matrices .
we remind that the asymptotic forms of @xmath39 and @xmath40 at large @xmath41 contain the plane waves and the spherical convergent waves , while the asymptotic forms of @xmath42 and @xmath43 at large @xmath41 contain the plane waves and the spherical divergent waves . in order to obtain the wave functions in the leading quasiclassical approximation and the first quasiclassical correction to it
, we exploit the convenient integral representation for the exact wave functions in the coulomb field suggested in ref .
the derivation of this representation is based on the relations @xmath44 and also @xmath45 where @xmath46 is the green function of the dirac equation in the coulomb field , @xmath47 and @xmath48 .
a convenient integral representation for @xmath46 was obtained in ref .
using eqs .
( 19)-(22 ) of that paper , we arrive at the following result for the wave functions @xmath49 and @xmath50 : @xmath51u_{\lambda{\bf p}}\,,\nonumber\\ & & v_{\lambda{\bf p}}^{(in)}({\bf r})=\frac{\exp(ipr)}{pr } \int\limits_0^\infty dt\,t^{2i\eta_p-1}{\mbox e}^{it^2}\nonumber\\ & & \times\left[s_b(x_{\bm p},\,pr)\left(t^2-z\alpha\frac{m}{p}\gamma^0\right ) ( 1-r_2)+is_a(x_{\bm p},\,pr)(1+r_2)\right]v_{\lambda{\bf p}}\,,\nonumber\\ & & s_a(x,\,\rho)=\sum_{l=1}^{\infty}{\mbox e}^{-i\pi\nu}j_{2\nu}(2t\sqrt{2\rho})\,l\frac{d}{dx}(p_l(x)+p_{l-1}(x))\ , , \nonumber\\ & & s_b(x,\,\rho)=\sum_{l=1}^{\infty}{\mbox e}^{-i\pi\nu}j_{2\nu}(2t\sqrt{2\rho})\frac{d}{dx}(p_l(x)-p_{l-1}(x))\quad , \nonumber\\ & & x_{\bm p}=\frac{\bm r\cdot \bm p}{rp}\ , , \quad r_{1,2}=(m\pm\gamma^0\varepsilon_p)\frac{\bm \gamma\cdot\bm r}{pr } \,.\end{aligned}\ ] ] here @xmath52 is the legendre polynomial , @xmath53 is the bessel function , @xmath54 . for the wave functions @xmath55 and @xmath56 we obtain : @xmath57\,,\nonumber\\ & & \bar v_{\lambda{\bf p}}^{(out)}({\bf r})=\frac{\exp(ipr)}{pr } \int\limits_0^\infty dt\,t^{2i\eta_p-1}{\mbox e}^{it^2}\nonumber\\ & & \times\bar v_{\lambda{\bf p}}\left[s_b(-x_{\bm p},\,pr ) ( 1+r_1)\left(t^2-z\alpha\frac{m}{p}\gamma^0\right)+is_a(-x_{\bm p},\,pr)(1-r_1)\right]\,.\end{aligned}\ ] ] the integrals over the variable @xmath58 in eqs .
( [ wf1 ] ) and ( [ wf2 ] ) can be expressed via the confluent hypergeometric functions . however , the forms ( [ wf1 ] ) and ( [ wf2 ] ) of the wave functions are more convenient for applications than the conventional ones . the results ( [ wf1 ] ) and ( [ wf2 ] ) are in agreement with the well - known solutions of the dirac equation in the coulomb field .
let us introduce the functions @xmath59 in terms of the functions ( [ fun123 ] ) , the wave functions @xmath50 and @xmath55 have the form @xmath60\,,\nonumber\\ & & { \cal r}_2^{(+)}=\sqrt{\frac{\varepsilon_p - m}{2\varepsilon_p}}\left[(\tilde f_b^{(+)}+\frac{z\alpha m}{p } f_b^{(+)})(\bm\sigma\cdot\hat{\bm p}+\bm\sigma\cdot\bm n)+f_a^{(+)}(\bm\sigma\cdot\hat{\bm p } -\bm\sigma\cdot\bm n)\right]\,,\nonumber\\ & & { \cal r}_1^{(-)}=\sqrt{\frac{\varepsilon_q+m}{2\varepsilon_q}}\left[(\tilde f_b^{(-)}+\frac{z\alpha m}{q } f_b^{(-)})(1+\bm\sigma\cdot\bm n\,\bm\sigma\cdot\hat{\bm q } ) + f_a^{(-)}(1-\bm\sigma\cdot\bm n\,\bm\sigma\cdot\hat{\bm q})\right]\,,\nonumber\\ & & { \cal r}_2^{(-)}=\sqrt{\frac{\varepsilon_q - m}{2\varepsilon_q}}\left[(\tilde f_b^{(-)}-\frac{z\alpha m}{q } f_b^{(-)})(\bm\sigma\cdot\hat{\bm q}+\bm\sigma\cdot\bm n)+f_a^{(-)}(\bm\sigma\cdot\hat{\bm q } -\bm\sigma\cdot\bm n)\right]\,,\end{aligned}\ ] ] where @xmath61 , @xmath62 , @xmath63 and @xmath64 then the matrix element @xmath30 , eq .
( [ m12 ] ) , is @xmath65\chi_{\lambda_2}\,.\end{aligned}\ ] ] for any vector @xmath66 we introduce the notation @xmath67 , @xmath68 , and write the matrix element @xmath30 in the form @xmath69\chi_{\lambda_2 } \,,\ ] ] where @xmath70 and @xmath71 are linear in @xmath72 , @xmath73 , @xmath74 and @xmath75 , while @xmath76 and @xmath77 are quadratic in these variables .
we have @xmath78\cdot\bm g^{(-)}\,,\quad \bm b_0=2\bm e\cdot \bm g^{(+)}\,\bm\nu+ 2 g\,\bm e\,,\nonumber\\ & & a_1=i[\bm\nu\times \bm e]\cdot(\bm\theta_p-\bm\theta_q)\ , g\,,\nonumber\\ & & \bm b_1=\bm e\cdot\bm g^{(+)}\,(\bm\theta_p+\bm\theta_q)+ \bm e\cdot(\bm\theta_p-\bm\theta_q)\,\bm g^{(-)}\nonumber\\ & & -(\bm\theta_p-\bm\theta_q)\cdot\bm g^{(-)}\,\bm e -\bm e\cdot(\bm\theta_p+\bm\theta_q)\,g\bm\nu\ , . \end{aligned}\ ] ] here @xmath79\nonumber\\ & & \pm\int d\bm r\,\exp(i\bm k\cdot\bm r)f_a^{(- ) } \left[(\bm n_{\perp}+\bm\theta_p){\tilde f}_b^{(+ ) } -\bm n_{\perp}{f}_a^{(+)}\right]\,,\nonumber\\ & & g=\frac{m\omega}{\varepsilon_p\varepsilon_q}\,\int d\bm r\,\exp(i\bm k\cdot\bm r)f_a^{(+ ) } f_a^{(-)}\,.\end{aligned}\ ] ] in ref .
@xcite , the following expressions for the sum @xmath80 and @xmath81 , which take into account the leading terms and first quasiclassical corrections , have been obtained @xmath82\,,\nonumber\\ & & s_b(x,\rho)=-\frac{y}{2\sqrt{2(1+x)}}j_1\left(y\sqrt{\frac{1+x}{2}}\right)\left[1+\frac{i\pi ( z\alpha)^2}{y}\right]\ , , \end{aligned}\ ] ] where @xmath83 .
these formulas are obtained for @xmath84 and @xmath85 .
substituting eq .
( [ sasb ] ) to eq .
( [ fun123 ] ) and taking the integrals over the variable @xmath58 we find @xmath86\,,\nonumber\\ & & f_b(\bm r,\,\bm p,\,\eta)=-\frac{i}{2}\exp\left(\frac{\pi\eta}{2}-i\bm p\cdot\bm r\right ) [ \gamma(1-i\eta)f(1+i\eta,\,2,\,i(pr+\bm p\cdot\bm r))\nonumber\\ & & + \frac{\pi\eta^2{\mbox e}^{i\frac{\pi}{4}}}{2\sqrt{2pr}}\gamma(1/2-i\eta)f(3/2+i\eta,\,2,\,i(pr+\bm p\cdot\bm r))]\,,\nonumber\\ & & { \tilde f}_b(\bm r,\,\bm p,\,\eta)=\frac{1}{2}\exp\left(\frac{\pi\eta}{2}-i\bm p\cdot\bm r\right ) [ \gamma(2-i\eta)f(i\eta,\,2,\,i(pr+\bm p\cdot\bm r))\nonumber\\ & & + \frac{\pi\eta^2{\mbox e}^{i\frac{\pi}{4}}}{2\sqrt{2pr}}\gamma(3/2-i\eta)f(1/2+i\eta,2,\,i(pr+\bm p\cdot\bm r))]\,.\end{aligned}\ ] ] here @xmath87 is the euler gamma function , and @xmath88 is the confluent hypergeometric function
. then we use the approach of ref .
@xcite based on the integral taken in ref .
@xcite , @xmath89 f(-ia_1,\,1,\ , i(qr+\bm q\cdot\bm r))\ , f(ia_2,\,1,\ , i(pr+\bm p\cdot\bm r ) ) \nonumber\\ & & = \frac{4\pi}{q^2}\left(\frac{m^2\omega ( 1+\xi_p\lambda)}{\varepsilon_p\xi_p q^2}\right)^{ia_1 } \left(\frac{m^2\omega ( 1+\xi_q\lambda)}{\varepsilon_q\xi_q q^2}\right)^{-ia_2 } f(-ia_1,ia_2,1,z)\,,\nonumber\\ & & \quad z=1-\frac{q^2\xi_p\xi_q(1+\lambda)}{m^2(1+\xi_p\lambda)(1+\xi_q\lambda)}\,,\quad \xi_p=\frac{1}{1+\delta_p^2}\,,\quad \xi_q=\frac{1}{1+\delta_q^2}\ , , \nonumber\\ & & \bm\delta_p=\frac{\varepsilon_p\bm\theta_p}{m}\,,\quad \bm\delta_q=\frac{\varepsilon_q\bm\theta_q}{m}\,,\quad \bm q=\bm p+\bm q-\bm k\,.\end{aligned}\ ] ] here we assume that @xmath90 . we write
@xmath91 and @xmath92 where the leading terms are @xmath93\,,\nonumber\\ & & \bm g^{(\pm)}_0=n\,\frac{(\varepsilon_p\mp\varepsilon_q)}{\omega}[(\xi_p\bm\delta_p+\xi_q\bm\delta_q)\,i\eta { \cal f } + ( \xi_p\bm\delta_p-\xi_q\bm\delta_q)\,(1-u){\cal f } ' ] \,,\nonumber\\ & & n =- i\frac{2\pi}{mq^2}\left(\frac{\varepsilon_q\xi_q}{\varepsilon_p\xi_p}\right)^{i\eta}|\gamma(1-i\eta)|^2\,,\nonumber\\ & & { \cal f}=f(-i\eta,\,i\eta,\,1,\ , u)\,,\quad { \cal f}'=\frac{\partial{\cal f}}{\partial u}\,,\quad u=1-\frac{\bm q^2_\perp}{m^2}\xi_p\xi_q\,.\end{aligned}\ ] ] here @xmath23 and @xmath94 is the hypergeometric function .
the first quasiclassical corrections , @xmath95 and @xmath96 , are given by the integrals @xmath97-(\bm p\leftrightarrow \bm q\,,\ , \eta\rightarrow -\eta)\bigg\}\,.\end{aligned}\ ] ] in order to take the integral over @xmath41 , we use the parameterization @xmath98 then , we obtain @xmath99 + ( \bm p\leftrightarrow \bm q\,,\ , \eta\rightarrow -\eta)\bigg\}\,,\nonumber\\ & & \delta\bm g^{(\pm)}=\frac{\pi^{3/2}\eta^2(\varepsilon_p\mp\varepsilon_q)}{2mq\omega}\left(\frac{\varepsilon_q\xi_q } { \varepsilon_p\xi_p}\right)^{i\eta}\int\limits_0^\infty\frac{d\lambda}{\sqrt{\lambda } } \left(\frac{1+\xi_p\lambda}{1+\xi_q\lambda}\right)^{i\eta}\nonumber\\ & & \times\bigg\{\frac{\sqrt{\xi_p}\gamma(1-i\eta)\gamma(1/2+i\eta)}{\varepsilon_q\sqrt{1+\xi_p\lambda } } \bigg[\bigg(-(1/2-i\eta)\frac{\xi_p\bm\delta_p}{1+\xi_p\lambda } + i\eta
\frac{\xi_q\bm\delta_q}{1+\xi_q\lambda}\bigg ) { \cal g}\nonumber\\ & & + \bigg(\frac{\xi_p\bm\delta_p}{1+\xi_p\lambda}-\frac{\xi_q\bm\delta_q}{1+\xi_q\lambda}\bigg)\ , ( 1-z){\cal g}'\bigg]- ( \bm p\leftrightarrow \bm q\,,\ , \eta\rightarrow -\eta)\bigg\}\,,\nonumber\\ & & { \cal g}=f(1/2-i\eta,\,i\eta,\,1,\ , z)\,,\quad { \cal g}'=\frac{\partial{\cal g}}{\partial z}\,.\end{aligned}\ ] ] here @xmath100 is defined in eq .
( [ nord ] ) .
using the matrix element obtained it is easy to write the cross section with all polarizations taken into account . for the cross section summed over the polarization of the electron and positron
, it is necessary to calculate @xmath101= 2[|a_0|^2+|\bm b_0|^2 + 2\mbox { re}(a_0a_1^*+\bm b_0\cdot\bm b_1^*)]\,,\ ] ] where we neglect the terms @xmath102 and @xmath103 .
it follows from eq .
( [ aabb ] ) that @xmath104 for any photon polarization .
thus , the terms with @xmath76 and @xmath77 do not contribute to the next - to - leading correction to the cross section summed over the electron and positron polarizations .
for simplicity , we restrict ourselves to the case of unpolarized photon . from eqs .
( [ aabb ] ) , ( [ m2 ] ) , ( [ gpmi ] ) , and ( [ deltag ] ) , we have @xmath105 , see eqs .
( [ gpmi ] ) and ( [ deltag ] ) .
we write @xmath106 , where the leading term is @xmath107\eta^2{\cal f}^2\nonumber\\ & & + [ u(\varepsilon_p^2+\varepsilon_q^2)+2\varepsilon_p\varepsilon_q(1-\xi_p-\xi_q)^2](1-u)^2{{\cal f}'}^2\bigg\}\,.\end{aligned}\ ] ] here @xmath108 and @xmath109 are defined in eq .
( [ gpmi ] ) .
the leading term is symmetric with respect to replacement @xmath110 .
the correction @xmath111 has the form @xmath112\nonumber\\ & & \times[4\varepsilon_p\varepsilon_q(\xi_pf_1+\xi_qf_2+f_3)+(\varepsilon_p^2+\varepsilon_q^2 ) ( f_1+f_2 + 2f_3)]\nonumber\\ & & + ( \varepsilon_p^2+\varepsilon_q^2)(1-u)[(f_1-f_2)i\eta{\cal f}-u(f_1+f_2){\cal f}']\,,\nonumber\\ & & f_1=\frac{(1/2-i\eta){\cal g}-(1-z){\cal g}'}{1+\xi_p\lambda}\,,\quad f_2=\frac{i\eta{\cal g}-(1-z){\cal g}'}{1+\xi_q\lambda}\,,\nonumber\\ & & f_3=\frac{(1-z){\cal g}'}{1+\lambda}\,.\end{aligned}\ ] ] here @xmath100 and @xmath113 are defined in eq .
( [ nord ] ) .
as should be , the correction @xmath111 is invariant under the replacement @xmath114 . since @xmath115 enters this expression only in the combination @xmath116
, it is evident that the correction @xmath111 is antisymmetric with respect to replacement @xmath117 , as well as with respect to replacement @xmath110 .
if @xmath118 , the leading term @xmath119 has the form @xmath120\,.\end{aligned}\ ] ] the correction eq .
( [ secnl ] ) at @xmath118 reads @xmath121 \nonumber\\ & & + ( \varepsilon_p-\varepsilon_q)\frac{(\varepsilon_p^2+\varepsilon_q^2)}{\varepsilon_p\varepsilon_q}\,\frac{q^2}{m^2 } \xi_p\xi_q\bigg\}\,.\end{aligned}\ ] ] in the limit @xmath122 and @xmath123 this formula reduces to @xmath124\nonumber\\ & = & -\,\frac{\alpha \eta^3(\varepsilon_p^2+\varepsilon_q^2 ) d\varepsilon_p\,d\bm\delta_p\,d\bm\delta_q}{\pi\delta_p^2\delta_q^2\omega^3q^3}(\bm\theta_q-\bm\theta_p)\cdot{\bm q}\ , \,.\end{aligned}\ ] ] the correction @xmath111 at @xmath125 , @xmath122 , and @xmath123 was also investigated in ref .
@xcite in scalar electrodynamics .
our result ( [ secabas ] ) , obtained for fermions , differs from the result of ref .
@xcite for scalar particles by the factor @xmath126 .
this factor is equal to @xmath127 for @xmath128 in accordance with the statement of ref .
@xcite . from the experimental point of view
, it may be interesting to consider the case @xmath129 or @xmath130 at @xmath131 .
then the leading symmetric term is @xmath132 where @xmath133 , @xmath134 , and @xmath135 .
this term is proportional to @xmath136 for any @xmath24 .
the leading antisymmetric term is @xmath137 it is also important to consider the asymtotics of the charge asymmetry in the region @xmath138 . in this case , the arguments @xmath108 and @xmath100 of the hypergeometric functions @xmath109 and @xmath113 , as well as the factor @xmath139^{i\eta}$ ] , in eq .
( [ secnl ] ) can be replaced by unity . as a result
, we find that @xmath140 and @xmath141 for any @xmath24 , and one can use eqs .
( [ secsb ] ) and ( [ secab ] ) for this region .
integration of eq .
( [ secs ] ) over @xmath142 gives for @xmath119 , ref .
@xcite , @xmath143\right\}\,,\nonumber\\ & & l=\ln\left(\frac{2\varepsilon_p\varepsilon_q}{m\omega}\right)-2-\mbox{re}[\psi(1+i\eta)+c]\,,\end{aligned}\ ] ] where @xmath144 is the euler constant .
the integration over @xmath142 gives the well - known result , ref .
@xcite , @xmath145 we have performed numerical integration of eq .
( [ secnl ] ) over @xmath146 .
[ dsadp ] shows the result of this integration , @xmath147 in units @xmath148 as a function of @xmath149 for a few values of @xmath150 . in units
@xmath148 on @xmath149 for a few values of @xmath150 : @xmath151 ( solid curves ) , @xmath152 ( dashed curves ) , and @xmath153 ( dotted curves ) ; @xmath154 ( tungsten ) . ] ( 0,0)(0,0 ) ( -110,-15 ) @xmath149 ( -240,50 ) the cross section @xmath111 integrated over both @xmath142 and @xmath146 was obtained in our paper ref .
@xcite and has the form @xmath155 the result of numerical integration of eq .
( [ secnl ] ) over @xmath142 and @xmath146 is in agreement with the above result .
in this section we estimate the contribution of the compton - type amplitude to the photoproduction cross section . since this amplitude is small in comparison with the leading amplitude found above , it is necessary to take into account only the interference between the compton - type amplitude and the leading amplitude .
the leading amplitude is enhanced at small angles @xmath156 and @xmath157 of the final particles .
therefore , it is sufficient to calculate the compton - type amplitude also at @xmath158 . for real initial and final photons with @xmath159 ( @xmath160 is the nuclear mass ) , the nuclear compton scattering amplitude , corresponding to the left diagram in fig.[fig : dia2 ] , in the forward direction reads @xmath161 where @xmath162 is the function measured for many nuclei , see review @xcite , @xmath163 and @xmath164 are the photon polarization vectors of the initial and final photons , respectively . the function @xmath162 satisfies the relations @xmath165=\frac{\omega^2}{2\pi^2}\,{\cal p}\!\!\int_0^\infty\frac{\sigma_{\gamma n}(\omega)}{\omega'^2-\omega^2 } \,d\omega'\,,\end{aligned}\ ] ] where @xmath166 is the nuclear photoabsorption cross section and @xmath167 denotes the integration in the principal value sense .
below pion photoproduction threshold , the cross section @xmath166 is conventionally written as a superposition of lorentzian lines @xmath168 where the parameters @xmath169 , @xmath170 , and @xmath171 are extracted from the experiment . the corresponding function @xmath162 has the form @xmath172 below pion threshold , but above the resonance region ( @xmath173 ) , the function @xmath162 has the form @xmath174 the last term in this asymptotics is equal to @xmath175 , where @xmath176 is the so called enhancement factor , see ref .
@xcite , and @xmath177 is the number of neutrons , and we obtain @xmath178 where @xmath179 is the proton mass , @xmath180 . for heavy nuclei @xmath181 , ref .
@xcite , so that @xmath182 using the function @xmath162 , we write the additional contribution to the photoproduction amplitude , corresponding to the right diagram in fig.[fig : dia2 ] , as follows @xmath183 we assume that @xmath184 and @xmath185 . taking the integral over @xmath186 we obtain @xmath187 \cdot\bm\gamma \,v _
{ \lambda_2\bm q}^{(in)}(\bm r ) \exp{(i\omega r ) } \,,\end{aligned}\ ] ] where @xmath61 . the main contribution to the integral over @xmath41 is given by the region @xmath188 and @xmath189 .
in this case @xmath190 using this asymptotics and taking the integral over @xmath41 we arrive at the compton - type correction to the photoproduction amplitude @xmath191\cdot\bm\vartheta\,,\quad \tilde{\bm b}=\tilde{n}\frac{(\varepsilon_p-\varepsilon_q)}{\omega } ( \bm e\cdot\bm\vartheta)\,\bm\nu\,,\nonumber\\ & & \tilde{n}=\,\frac{2\pi\,t(\omega)}{m\omega}\left(\frac{\varepsilon_q } { \varepsilon_p}\right)^{i\eta } \frac{1}{1+\vartheta^2}\,,\nonumber\\ & & \bm\vartheta=\frac{\varepsilon_p\varepsilon_q}{m\omega}(\bm\theta_p-\bm\theta_q)= \frac{\varepsilon_q\bm\delta_p-\varepsilon_p\bm\delta_q}{\omega}\,.\end{aligned}\ ] ] the corresponding correction to the cross section has the form @xmath192\,.\end{aligned}\ ] ] this correction contains both symmetric and antisymmetric parts with respect to replacement @xmath193 .
the symmetric part is proportional to @xmath194 and the antisymmetric part is proportional to @xmath195 @xmath196\bigg)\,,\nonumber\\ & & \mu = \eta\ln\left(\frac{\xi_q}{\xi_p}\right)\,.\end{aligned}\ ] ]
in quantum electrodynamics , an electron differs from a positron only by its charge .
therefore , the cross section of @xmath15 pair photoproduction satisfies the relation @xmath197 we define the charge asymmetry @xmath0 as @xmath198 let us first neglect @xmath199 and calculate @xmath200 as a ratio of @xmath111 in eq .
( [ secnl ] ) and @xmath119 in eq .
( [ secs ] ) . outside the very narrow region @xmath201
, we can replace @xmath202 .
then , at fixed @xmath203 , @xmath204 , and @xmath150 the asymmetry @xmath205 scales as @xmath3 , as can be seen from eqs .
( [ secnl ] ) and ( [ secs ] ) .
figure [ sigmadd ] shows the dependence of @xmath0 on @xmath149 in units @xmath3 for tungsten ( @xmath154 ) for a few values of @xmath206 and @xmath207 , where @xmath207 is the angle between vectors @xmath142 and @xmath146 .
it is seen that the charge asymmetry may be rather large ( @xmath208 for @xmath209 ) .
the asymmetry is large when @xmath149 and/or @xmath206 are much larger than unity .
note that this statement is also valid in the region @xmath138 ( but @xmath210 and @xmath123 ) .
on @xmath149 in units @xmath3 for a few values of @xmath206 , @xmath207 , and @xmath150 : @xmath151 ( solid curves ) , @xmath152 ( dashed curves ) , and @xmath153 ( dotted curves ) , @xmath154 ( tungsten ) . ]
( 0,0)(0,0 ) ( -205,-15)@xmath149 ( -400,180 ) it is interesting to understand the importance of high - order in @xmath24 terms in the charge asymmetry .
figure [ zadependence ] shows the dependence of @xmath0 on @xmath23 in units @xmath3 for @xmath211 , @xmath212 and a few values of @xmath150 and @xmath207 .
the dashed curve in this figure is obtained in the leading in @xmath24 approximation ( linear in @xmath24 ) .
it is seen that @xmath24 dependence is very strong even for intermediate values of @xmath24 .
on @xmath23 in units @xmath3 for @xmath211 , @xmath212 and a few values of @xmath213 and @xmath207 .
solid curve represents the exact in @xmath24 result , dashed curve is obtained in the leading in @xmath24 approximation ( linear in @xmath24 ) . ] ( 0,0)(0,0 ) ( -200,-15)@xmath24 ( -410,180 ) let us also introduce the charge asymmetry @xmath214 for the cross section integrated over the angles of one of the particles , @xmath215 we calculate nominator in @xmath214 integrating @xmath111 in eq .
( [ secnl ] ) over @xmath146 and denominator using eq .
( [ secsdeltap ] ) .
figure [ sigmadp ] shows the dependence of @xmath214 on @xmath149 for @xmath154 ( tungsten ) , @xmath209 , and a few values of @xmath213 .
note that @xmath214 , in contrast to @xmath0 , does not scale as @xmath3 due to logarithmic dependence of @xmath216 on @xmath4 , see eq .
( [ secsdeltap ] ) .
it is seen that @xmath214 is noticeable though it is smaller than @xmath0 .
on @xmath149 for @xmath151 ( solid curves ) , @xmath152 ( dashed curves ) , and @xmath153 ( dotted curves ) ; @xmath154 ( tungsten ) , @xmath209 . ] ( 0,0)(0,0 ) ( -110,-15 ) @xmath149 ( -220,60 ) the charge asymmetry corresponding to the cross section integrated over the angles of both particles ( over @xmath142 and @xmath146 ) is very small ( see ref .
@xcite ) . let us discuss now the contribution @xmath2 to the charge asymmetry , @xmath217 where @xmath218 is given by eq .
( [ seccomptona ] ) and @xmath219 is given by eq .
( [ secs ] ) . in the region @xmath220 , we have @xmath221 and @xmath222 . in this region
, @xmath199 may be comparable with @xmath111 only for light nucleus ( @xmath125 ) , where asymmetry is very small . in the region
@xmath223 , @xmath224 , we have @xmath221 and @xmath225 .
again , @xmath199 may be comparable with @xmath111 only for light nucleus where asymmetry is very small .
the only region where @xmath226 and @xmath2 is not too small is @xmath125 , @xmath122 , @xmath123 , but @xmath227 . in this region
the ratio @xmath228 is @xmath229 and may be larger than unity .
here we took into account that @xmath230 but @xmath231 .
for the contribution @xmath2 to the charge asymmetry we have @xmath232 where @xmath233 , so that @xmath2 may reach about ten percent at large transverse momenta compared to the electron mass , see fig .
[ fig : tildea ] . as a function of @xmath149 for @xmath234 , @xmath235 ( proton ) , @xmath236 , @xmath237 . ]
( 0,0)(0,0 ) ( -2.5,-0.3 ) @xmath149 ( -5.4,1.5 )
we have derived exactly in the parameter @xmath23 the charge asymmetry @xmath0 , eq .
( [ asr ] ) , in the process of @xmath15 photoproduction in a coulomb field at photon energy @xmath11 , @xmath238 , and @xmath239 .
this asymmetry is related to the first quasiclassical correction to the differential cross section of the process , eq .
( [ secnl ] ) . when @xmath240 and/or @xmath241 are much larger than the electron mass @xmath1 , the charge asymmetry can be as large as tens percent .
the charge asymmetry @xmath214 , eq . ( [ asr1 ] ) , in the cross section integrated over the transverse momenta of one of the particles is several times smaller than the asymmetry @xmath205 in the cross section differential with respect to the transverse momenta of both particles .
we have also estimated the contribution @xmath2 , eqs .
( [ seccomptona ] ) and ( [ asrcompton ] ) , to the charge asymmetry of the compton - type diagram . for @xmath242 ,
this contribution is negligible .
the only region where @xmath2 can be important is @xmath125 ( light nucleus ) , @xmath230 but @xmath231 .
though we have performed our calculation for the pure coulomb field , our results are also applicable for photoproduction in the electric field of atoms except the very narrow region @xmath243 , where @xmath244 is the atomic screening radius .
our results clearly demonstrate that experimental observation of the charge asymmetry in the process of @xmath15 photoproduction in a coulomb field is a realistic task .
we thank s.j . brodsky and g. ron for valuable discussions . this work was supported in part by the rfbr grant no . 09 - 02 - 00024 and
the grant 14.740.11.0082 of federal special - purpose program `` scientific and scientific - pedagogical personnel of innovative russia '' .
v. b.berestetski , e. m.lifshits , l. p.pitayevsky , _ quantum electrodynamics _
( pergamon , oxford , 1982 ) .
milstein , i.s .
terekhov , zh .
. fiz . * 125 * , 785 ( 2004 ) [ jetp * 98 * , 687 ( 2004 ) ] .
milstein , v.m .
strakhovenko , phys .
a * 90 * ( 1982 ) 447 .
h. olsen and l.c .
maximon , phys . rev . * 114 * , 887 ( 1959 ) .
a. nordsieck , phys . rev . * 93 * , 785 ( 1954 ) .
|
first quasiclassical correction to the differential cross section of high - energy electron - positron photoproduction in the electric field of a heavy atom is obtained with the exact account of the field .
this correction is responsible for the charge asymmetry @xmath0 in this process .
when the transverse momentum of at least one of the produced particles is much larger than the electron mass @xmath1 , the charge asymmetry can be as large as tens percent .
we also estimate the contribution @xmath2 to the charge asymmetry coming from the compton - type diagram . for heavy nuclei ,
this contribution is negligible . for light nuclei ,
@xmath2 is noticeable only when the angle between the momenta of electron and positron is of order of @xmath3 ( @xmath4 is the photon energy ) while the transverse momenta of both particles are much larger than @xmath1 .
|
the decays of @xmath4 lepton provide unique opportunities to search for the effects beyond the standard model ( bsm ) @xcite .
the large mass of @xmath4 ( @xmath5 mev/@xmath6 @xcite ) , in comparison with that of the electron or muon , allows one to expect an essential enhancement in the sensitivity to the effects of new physics ( np ) @xcite . of all tau decays , the leptonic ones are precisely calculated within electroweak sector of the sm ,
hence they offer a clean laboratory to search for the effects of np . through the measurement of michel parameters , @xmath7 , @xmath8 , @xmath9 and @xmath10 , in ordinary leptonic decays @xmath11 ( @xmath12 ) ,
the experimental verification of the lorentz structure of the charged weak interaction was carried out @xcite .
the most precision measurements were done by aleph @xcite and cleo @xcite collaborations .
the measurement of the ratio of the branching fractions @xmath13 is used to test the lepton universality : @xmath14 where @xmath15 and @xmath16 are the couplings of @xmath4 with @xmath17 and @xmath18 , respectively , @xmath19 ( @xmath12 ) is outgoing lepton mass , and @xmath20 is a known function @xcite .
the most precise measurement at @xmath21 , @xmath22 @xcite , is consistent with lepton universality .
radiative leptonic decay @xmath23 ( unless specified otherwise , charge - conjugated decays are implied throughout the paper ) provides an additional promising tool to search for np .
feynman diagrams of this decay in the sm are presented in fig [ rd ] .
the presence of the radiation exposes the internal structure of @xmath4 decays differently from the ordinary leptonic decays .
for instance , the measurement of the spectra of outgoing lepton and photon allows one to access three more michel parameters , @xmath24 , @xmath25 and @xmath26 @xcite . in this note
we consider the anomalous four - point @xmath0 scalar and tensor couplings .
externalph1 ( 120,60 ) externalph2 ( 120,60 ) externalph3 ( 120,60 )
we suggest to consider anomalous four - point @xmath0 scalar and tensor interactions ( see fig . [ fptidiagram ] ) , the modified lagrangian of the charged weak interaction of @xmath4 is written as : @xmath27 , \label{formla_np}\end{aligned}\ ] ] where the first term is the sm lagrangian , @xmath28 and @xmath29 characterize the magnitudes of the scalar and tensor interactions , respectively , @xmath30 is the electromagnetic field , @xmath31/2 $ ] , @xmath18 is electron charge .
the introduced terms appear from the gauge invariant dimension - five operators @xmath32 and @xmath33 \psi(\tau)$ ] , where @xmath34 is the @xmath35 gauge covariant derivative , @xmath36\equiv d^\mu d^\nu - d^\nu d^\mu$ ] .
anomolus ( 120,60 ) the total matrix element squared of the @xmath37 decay in the presence of the anomalous amplitude @xmath38 is written as : @xmath39 the contribution to the differential decay width from the @xmath40 is given by : @xmath41 , \label{dff}\end{aligned}\ ] ] where @xmath42 gev@xmath43@xcite is the fermi constant and @xmath44@xcite is the fine - structure constant , @xmath45 and @xmath46 are four - momenta of the outgoing charged lepton and photon , respectively ; @xmath47 , @xmath48 and @xmath49 are the solid angles of the final charged lepton and photon , respectively ; @xmath50 , @xmath51 , @xmath52 , @xmath53 ( asterisks indicate parameters measured in the @xmath4 rest frame ) and @xmath54 ; @xmath55 and @xmath56 are the form factors ( see appendix for the explicit formulae ) . integrating the differential decay width numerically ,
we obtain : @xmath57 , \nonumber\end{aligned}\ ] ] where the coefficients @xmath58 and @xmath59 @xmath60 or @xmath61 are : @xmath62 , @xmath63 , @xmath64 , @xmath65 , @xmath66 , @xmath67 , @xmath68 and @xmath69 , where the error is statistical uncertainty of numerical integration . in this calculation
, we use the photon energy threshold of @xmath70 in the @xmath4 rest frame .
taking into account the sm prediction @xmath71 @xcite and the pdg average @xmath72 @xcite , we get the constraint at 95% cl : @xmath73 , or : @xmath74 moreover , the four - point @xmath0 scalar and tensor interactions can be searched for at the electron - positron colliders , in the process of the @xmath75 annihilation @xmath76 .
this anomalous coupling is responsible for the production of the single @xmath4 lepton below @xmath77 production threshold .
such a mechanism can be searched for at the low energy @xmath75 colliders like beijing electron - positron collider ( bepc ) , cornell electron storage ring ( cesr ) and vepp-2000 @xcite , as well as at the b - factories , belle @xcite / kekb @xcite and @xmath21 @xcite / pep - ii , in the processes of @xmath75 annihilation with the initial state radiation @xcite .
for example , at vepp-2000 in the center - of - mass energy range from about @xmath78 gev up to @xmath79 gev ( near the production threshold of single tau ) , @xmath80 lepton is produced almost at rest , accompanied with @xmath81 , @xmath82 or @xmath83 .
as a result , besides @xmath84 events , a clear signature of the production of single @xmath80 will be the monochromatic @xmath85 or @xmath86 ( from @xmath87 or @xmath88 decays ) .
the precision measurement of the properties of leptonic decays of @xmath4 lepton offers unique opportunity to search for the physics beyond the standard model .
the radiative leptonic decay @xmath89 provides an additional tool to probe the internal structure of the weak interaction .
the anomalous four - point @xmath0 scalar and tensor interactions are simple extensions of the standard model , which affect the spectra of the daughter particles in the radiative leptonic decays of tau .
we calculated the corresponding differential and the total decay widths .
the world average value of the branching ratio of the @xmath1 decay and its sm prediction constrain the magnitudes of the scalar and tensor couplings to be @xmath2 and @xmath3 ( at 95% cl ) , respectively . in this note
, we extract the constraints on the @xmath90 and @xmath91 coupling constants from the total widths , but the more sensitive method is to fit the full differential decay width of the radiative leptonic decay of tau .
we are grateful to andrey pomeransky ( budker institute of nuclear physics ) for the fruitful discussions . 99 a. stahl and h. voss , z. phys .
c*74 * , 73 ( 1997 ) .
a. gabriel _ et al .
phys . b*582 * 3 ( 2000 ) . g. l. castro and n. quintero , phys .
rev . d*85 * , 076006 ( 2012 ) . c. deb _ et al .
rev . d*85 * , 011301 ( 2012 ) .
j. fan _ et al .
_ , jhep * 01 * 111 ( 2016 ) .
a. moyoti and g. t. velasco , phys .
rev . d*86 * , 013014 ( 2012 ) .
j. p. lees _ et al .
_ , ( the babar collaboration ) phys .
rev . lett . * 114 * , 171801 ( 2015 ) .
h. albrecht _
_ , z. phys .
c*68 * 25 ( 1995 ) .
a. brignole and a. rossi , nucl .
phys . b*701 * 3 ( 2004 ) .
l. calibbi _
et al . _ ,
phys . rev . d*74 * 116002 ( 2006 ) .
et al . _ , phys .
rev . d*66 * 115013 ( 2002 ) .
k. a. olive , _ et al .
_ , ( particle data group ) , chin . phys
. c*38 * , 090001 ( 2014 ) .
s. eidelman and m. passera , mod .
a*22 * , 159 ( 2007 ) .
a. heister _ et al .
_ , eur .
j. c*22 * , 217 ( 2001 ) .
alexander _ et al .
_ , phys .
d*56 * , 5320 ( 1997 ) .
w. j. marciano and a. sirlin phys .
lett . * 61 * , 1815 ( 1988 ) .
b. aubert _ et al .
_ , ( the babar collaboration ) phys . rev
. lett .
* 105 * , 051602 ( 2010 ) .
m. l. perl , arxiv:9812400 ( 1998 ) .
c. fronsdal , h. uberall , phys .
* 113 * 654 ( 1939 ) .
a. b. arbuzov , _ et al .
_ , arxiv:1605.06612 ( 2016 ) .
n. shimizu _ et al .
_ , nucl .
instrum .
a*824 * , 11 , 237 ( 2016 ) . 0 j. polak and m.zralek , phys . rev .
d*46 * , 3871 ( 1992 ) .
a. stahl , nucl .
b*76 * , 173 ( 1999 ) .
r. prieels_et al . _ ,
d*90 * , 112003 , ( 2014 ) .
p. abreu , _
et al . _ ,
j. c*16 * , 229 ( 2000 ) .
t. bergfeld , _ et al .
_ , phys .
lett . * 84 * , 830 ( 2000 ) .
j. p. lees , _ et al .
_ phys .
d * 91 * , 051103 .
t. kinoshita and a. sirlin , phys . rev . * 113 * , 6 ( 1958 ) . t. kinoshita and a. sirlin , phys .
* 2 * , 177 ( 1959 ) .
m. fael , _ et al . _ , jhep . * 07*. 153 j ( 2015 ) .
m. shatunov et al . , conf .
c * 0006262 * , 439 ( 2000 ) .
d. e. berkaev et al . , nucl .
* 225 - 227 * , 303 ( 2012 ) .
a. abashian _ et al .
_ ( belle collaboration ) , nucl .
methods phys .
a * 479 * , 117 ( 2002 ) ; also see detector section in j.brodzicka _ et al . _ ,
2012 * , 04d001 ( 2012 ) .
s. kurokawa and e. kikutani , nucl .
methods phys .
a * 499 * , 1 ( 2003 ) , and other papers included in this volume ;
t.abe _ et al .
. phys . * 2013 * , 03a001 ( 2013 ) , and following articles up to 03a011 .
a. j. bevan _ et al .
_ [ babar and belle collaborations ] , eur .
j. c * 74 * , 3026 ( 2014 ) [ arxiv:1406.6311 [ hep - ex ] ] .
b. aubert _ et al .
_ [ babar collaboration ] , nucl .
instrum .
a * 479 * , 1 ( 2002 ) [ hep - ex/0105044 ] .
|
as one of the extensions of the standard model , we investigate the anomalous four - point @xmath0 scalar- and tensor - type interactions , which originate from the gauge invariant dimension - five operators .
the coupling constants are constrained by the measured branching ratio of the @xmath1 decay : @xmath2 and @xmath3 at the 95% confidence level for the scalar and tensor interactions , respectively .
|
soon after the discovery of the compact radio source sagittarius a * ( sgr a * ) at the galactic center ( gc ) @xcite , davies , walsh & booth ( 1976 , hereafter dwb ) carried out observations of sgr a * at 0.408 , 0.96 and 1.66 ghz .
sgr a * was only detected at the two higher frequencies .
the measurements showed a low - frequency turnover around 1 ghz in the spectrum of sgr a*. the authors attributed the decrease in flux density below 1 ghz to free - free absorption from ionized gas of sgr a west .
recent observations of sgr a * at both 620 and 330 mhz indicate that the spectrum below 1 ghz may indeed have a pronounced turnover . at 620 mhz ,
roy & rao ( 2004 ) detected sgr a * using the gmrt with a flux density of @xmath3 jy with an angular resolution of @xmath4 , while @xcite detected sgr a * at 330 mhz using the vla of the national radio astronomy observatory ( nrao ) with a flux density of @xmath5 jy ( resolution of @xmath29 ) .
sgr a * is now believed to be associated with a supermassive black hole ( smbh ) with a mass @xmath6 @xcite .
a number of models have been proposed to explain the observed phenomena in the vicinity of sgr a * at radio , near - ir and x - ray wavelengths ( melia & falcke 2001 ) .
coker & melia ( 1997 ) proposed that stellar winds , which originate from massive stars near sgr a * , might play an important role in fuelling sgr a * and therefore may have caused the variations in flux density at radio wavelengths .
these models are sensitive to the observed spectrum of sgr a*. simultaneous multi - wavelength observations could provide important constraints for current accretion or outflow theories . if stellar winds are indeed responsible for the flux density variations at millimeter / submillimeter wavelengths , a cutoff at long radio wavelengths could be sensitive to changes in stellar wind flux near sgr a*. to determine the low - frequency turnover in the spectrum of sgr a * , we have performed multi - wavelength observations using the vla at 90 , 26 , 23 , 20 , 17 , 6 , 3.6 , 2 , 1.3 and 0.7 cm , and the gmrt at 47 cm on 17 june 2003 ( ut ) . at the same epoch , both the sma ( moran 1998 ; ho , moran & lo 2004 ) and the keck ii 10 m telescope observed sgr a * at 0.89 mm and 3.8 @xmath0 m , respectively . in this letter , we present results from the quasi - simultaneous observations .
the vla and gmrt observations are summarized in table [ tab : obs ] , including observing band ( @xmath7 ) , frequency ( @xmath8 ) , bandwidth ( @xmath9 ) , on - source time ( @xmath10 ) and flux density of the primary calibrator ( @xmath11 ) .
the observations with the vla were carried out in the a configuration at wavelengths from 90 cm to 7 mm .
the observations at 90 and 20 cm with the vla and at 47 cm with the gmrt were performed in spectral - line modes in order to reject radio frequency interference and minimize bandwidth smearing .
the raw data were reduced using the astronomical image processing system ( aips ) of the nrao following standard procedures .
absolute flux density calibration was performed using 3c 286 ( at 0.7 , 1.3 and 2 cm ) or 3c 48 ( at other wavelengths ) following the flux density scales of baars et al .
the flux density calibration of 3c 48 or 3c 286 was carried out using the visibilities over appropriate _
uv _ ranges suggested by vla documentation at wavelengths between 90 and 2 cm ; clean models of 3c 286 ( obtained from c.chandler ) were used for flux density calibrations at 1.3 and 0.7 cm using all visibilites .
a few nearby qsos were interleaved during the observations of sgr a * to determine the complex gains .
an amplitude and phase calibration of 3c 48 or 3c 286 was performed using a two - minute solution interval to obtain average flux density scales over the scan , which were then applied to the phase calibrators and sgr a*. a few iterations of phase self - calibrations of calibrated _ uv _ data were also performed to correct for residual phase errors . the gmrt calibration procedure for observation of the galactic center
is described in detail by roy & rao ( 2004 ) .
the primary flux density calibrator was 3c 48 , whose flux density is 28.5 jy at 640 mhz ( baars scale ) .
corrections were made for the increase of the system temperature in the direction of the galactic center compared to the high galactic latitude field of 3c 48 .
the point sources 1751@xmath12253 and 1741@xmath1225 were used as secondary phase and bandpass calibrators , respectively .
the flux densities of sgr a * were determined in the image domain . either a point - source or an elliptical gaussian model was used to fit the data .
the flux density measurement at 90 cm will be discussed in section [ detection ] . at 20 cm ,
sgr a * is slightly resolved ; thus a 2-d elliptical gaussian function was used to fit the long baseline ( @xmath1320 k@xmath7 ) visibilities . at wavelengths between 6 and 0.7 cm , free - free emission from the hii regions in sgr a west
is resolved out on baselines longer than 80 k@xmath7 .
the flux density of a point source can be directly determined in the _ uv _ domain by averaging the amplitude of the visibilities on baselines @xmath14 .
the difference in flux densities between the two procedures is consistent with the uncertainty ( @xmath15 ) in the measurements .
the calibration uncertainty ( @xmath16 ) is dominated by the uncertainty in the calibration of the flux density scale .
the final 1@xmath17 errors are the quadrature addition of the two terms , _
i.e. _ @xmath18 .
the measurements of flux densities at 2 , 1.3 and 0.7 cm are consistent with the values derived from the vla monitoring program @xcite .
the gmrt image at 47 cm was made with a _
uv _ range 1050 k@xmath7 to minimize confusion from the sgr a complex .
a gaussian model was fitted to the flux density distribution of sgr a * and a constant term with a slope was fitted to the background using jmfit in aips .
the flux density of sgr a * is [email protected] jy at 47 cm .
figure [ fig : sgra90 ] shows the central @xmath20 region of the galaxy at 90 cm .
the supernova sgr a east dominates the total flux density at 90 cm .
an excess in flux density is observed at the expected location of sgr a * ( red cross at the image center in figure [ fig : sgra90 ] ) ; the region surrounding sgr a * is significantly absorbed by the ionized gas in sgr a west , except for the extended emission south of sgr a*. we made two slices along the major ( pa@xmath21 ) and minor ( pa@xmath22 ) axes of the expected scattering shape to determine the flux density of sgr a * ( figure [ fig : slice ] ) .
gaussian profiles were fitted to each of the slices after subtracting diffuse emission from the background ; an apparent size of @xmath23 was determined .
the errors in the parameters are determined using the 2-d gaussian fitting procedure jmfit in aips and also checked by fitting the two 1-d slices ( figure [ fig : slice ] ) .
the source size of sgr a * at 90 cm is @xmath24 and @xmath25 , after deconvolution .
the fitted peak flux density is @xmath26 mjy / beam , with a total flux density of @xmath27 mjy .
the observed flux density and source size of sgr a * at 90 cm agree with the determinations of @xcite within 1@xmath17 .
figure [ fig : spec ] shows the quasi - simultaneous spectrum of sgr a * at wavelengths ranging from 90 cm to 3.8 @xmath0 m , observed on 17 june 2003 .
table [ tab : spec ] summarizes the flux density measurements .
a rising spectrum extends from @xmath22 cm to the submillimeter band .
however , the wavelength of the peak emission is quite uncertain since there are no measurements from 0.89 mm to 3.8 @xmath0 m .
the rising spectrum from the short cm to mm wavelengths can be described by a power law ( @xmath28 ) with @xmath29 ( _ solid line _ in figure [ fig : spec ] ) ; the spectrum between 3.6 and 47 cm is flatter , @xmath30 ( _ dashed line _ in figure [ fig : spec ] ) .
the difference in spectral indices between the two parts of the power - law spectra is @xmath31 , suggesting a significant break near 3.6 cm on 17 june 2003 .
a break in the cm - band spectrum of sgr a * has been suggested previously @xcite
. the determined break wavelength of 34 cm is consistent with the previously suggested break of 23.5 cm suggested by falcke et al .
the excess in flux density towards short wavelengths in the spectrum of sgr a * might be associated with activities in the inner region of the accretion disk or jet nozzle ( melia & falcke 2001 ) . in comparison with the radio light curves ,
the increased spectral index at wavelengths shorter than the break of @xmath23.6 cm on 17 june 2003 is consistent with the fact that sgr a * was undergoing a large radio outburst .
the vla monitoring campaign ( herrnstein et al . 2004 ) did show that sgr a * was in an outburst stage at 7 mm near the epoch of 16 june 2003 ; the outburst apparently peaked near the epoch of 17 july 2003 .
the flux density at 7 mm on 17 july 2003 was at the highest level between 21 june 2000 and 16 october 2003 ( herrnstein et al . 2004 ) .
therefore , a break in the spectrum of sgr a * at a short cm wavelength could be a signature of synchrotron self - absorption @xcite of a relatively strong radio outburst , as proposed to explain the 3 october 2002 x - ray event and corresponding radio outburst @xcite . at 90 cm ,
the measured flux density of [email protected] jy is far below the inferred 0.49 jy extrapolated from the power law fit @xmath32 .
this significant decrease ( @xmath33 ) in flux density suggests that the spectrum of sgr a * on 17 june 2003 could have a cutoff at wavelengths longer than 47 cm .
the earlier jodrell bank measurements of sgr a * made by dwb showed a low - frequency turnover at about 1 ghz .
the flux densities of [email protected] jy at 0.96 ghz and [email protected] jy at 1.66 ghz had been corrected for resolutions of the interferometer and the known angular diameters of sgr a*. we note that the angular diameters measured with the jodrell bank at 0.96 and 1.66 ghz are consistent with the values extrapolated from the relationship between the angular size and observing wavelength ( _ e.g. _ , lo et al .
1998 ; bower et al . 2004 ) . for angular sizes of @xmath34 at 0.96 ghz and @xmath35 at 1.66 ghz ,
sgr a * is dominant on baselines from 34 to 75 k@xmath7 and from 60 to 132 k@xmath7 , respectively .
the errors in flux densities at these two frequencies ( @xmath210% ) are consistent with those estimated from our current data at 20 cm . however , the 408 mhz upper limit measured by dwb with a single baseline is quite uncertain because of confusion arising from the complex region in sgr a east in close proximity to the broadened source sgr a * at 408 mhz ( _ e.g. _ @xmath28 , roy & rao 2004 ) .
this upper limit will be excluded in following discussion .
figure [ fig : lfspec ] shows the observed spectra for sgr a * at epochs 1975 ( dwb ) and 2003 ( this letter ) for wavelengths longer than 3.6 cm . a ratio of @xmath36 ( @xmath37(2003)/@xmath38(1975 ) ) would suggest a significant increase in the flux density at about 1 ghz from epoch 1975 to 2003 .
the spectra at both epochs can be fitted with a model of a slowly rising power law spectrum ( @xmath39 ) together with a free - free absorption screen ( @xmath40 ) between sgr a * and the observer , as first proposed by dwb . if the flux density at 0.96 ghz measured by dwb is indeed reliable , then the comparison of the fits to the 1975 ( _ dashed line _ in figure [ fig : lfspec ] ) and 2003 data ( _ solid line _ in figure [ fig : lfspec ] ) does suggest a substantial change in the spectrum of sgr a * at wavelengths longer than 30 cm in the past 28 years . if the free - free absorption model is in fact correct , the variation in the low - frequency spectrum would suggest that the column density of the free - free absorption screen must have decreased significantly over the past 28 years . starting from dwb
s free - free absorption model ( @xmath41 ) , the quantitative change in the column density of ionized gas in front of sgr a * can be assessed .
the data observed at both epochs can be fitted with a power - law spectrum @xmath42 at the higher frequencies along with an exponential cutoff at the lower frequencies arising from free - free absorption .
the comparison of the model fits to the observations at these two epochs suggests that free - free absorption opacity ( @xmath43 towards sgr a * must have decreased by a factor of nine from 1975 to 2003 .
the critical cutoff in wavelength due to free - free absorption has changed from @xmath230 cm to @xmath2100 cm . assuming the electron temperature of the ionized gas remains constant , the inferred decrease in @xmath44 would then correspond to a decrease by a factor of nine in emission measure @xmath45 .
such a large variation in em is unlikely to arise from changes in electron density in the halo of sgr a east ( pedlar et al .
if the changes in em and electron density do occur in a compact region within 0.110 of sgra * ( 0.0040.4 pc ) , stellar winds which originate from massive stars towards sgr a * could provide the changing environment .
for example , loeb ( 2004 ) suggested that stellar winds from the close - in stars could modulate low - frequency flux density of sgr a * over a timescale of about ten years through orbital motions of the stars .
here we consider two types of stellar winds : ( 1 ) the stellar wind is isotropic ( @xmath46 ) and ( 2 ) the stellar wind forms a one - dimensional flow due to the gravity of the smbh ( @xmath47 ) . using the star s2 @xcite as an example , the orbit period is about 15 years and the ratio of the pericenter and apocenter is about 1:15 ; the star has a surface temperature of about 30000k , providing an ionization zone of h roughly equivalent of a b0 star ( ghez et al .
the stellar wind mass flux at pericenter in case ( 1 ) is about 200 times that at apocenter ; in case ( 2 ) the ratio in stellar wind flux between pericenter and apocenter is 15:1 .
the ratio of three in electron column density between 1975 and 2003 inferred from the change of the low - frequency cutoff is much less than that of the stellar wind flux in both cases .
nevertheless , large amplitude fluctuations of accreted matter within 0.002 pc ( 0.05 ) captured from the stellar winds near sgr a * over a time scale of less than a few decades were predicted in numerical simulations ( melia & coker 1999 ) .
a variation in the radio continuum spectrum was also suggested by the simulations , although the authors attributed the low - frequency turnover to refraction effects .
it is more likely that the source for the low - frequency turnover is free - free absorption from cool gas in close proximity to sgr a*. the colder gas ( @xmath48 k ) can co - exist with the hot gas ( @xmath49 k ) that provides the x - ray emission near sgr a * ( cuadra et al .
2005a , 2005b ) . for a stellar wind with a size of 0.004 pc ,
the free - free absorption opacity is about unity at about 0.4 ghz for an electron density of @xmath50 @xmath51 and electron temperature of @xmath50 k. the variation of the low - frequency turnover due to free - free absorption towards sgr a * appears to be sensitive to fluctuations of the stellar wind mass captured by the smbh potential .
we thank the referee and h. falcke for helpful comments .
the work is supported by the nsfc ( 10503008 , 10328306 , 10333020 ) .
the vla is operated by the national radio astronomy observatory , a facility of the national science foundation operated under cooperative agreement by associated universities , inc .
the gmrt is a component of the national centre for radio astrophysics of the tata institute of fundamental research .
mjy / beam . in order to compare our results with those obtained by @xcite at the same wavelength
, we have restored the final image using a similar synthesized beam , _
i.e. _ , @xmath52 ( pa@xmath22 ) . the _ r.m.s . _ noise in the image is 12 mjy / beam .
the red cross at the image center marks the expected scattering size and position of sgr a * at 90 cm , _
i.e. _ , @xmath2@xmath53 at pa= 80 @xcite . , scaledwidth=40.0% ] [email protected]@xmath55 , @xmath56 ( b1950 ) .
_ left : _ slice through the major axis ( pa@xmath21 ) with a resolution of 6.8 .
east is to the left ; _ right : _
slice through the minor axis ( pa@xmath22 ) with a resolution of 10.9 .
south is to the left .
, scaledwidth=50.0% ]
|
we observed sgr a * using the very large array ( vla ) and the giant metrewave radio telescope ( gmrt ) at multiple cm and mm wavelengths on 17 june 2003 .
the measured flux densities of sgr a * , together with those obtained from the submillimeter array ( sma ) and the keck ii 10 m telescope on the same date , are used to construct a simultaneous spectrum of sgr a * from 90 cm to 3.8 @xmath0 m .
the simultaneous spectrum shows a spectral break at about 3.6 cm , a possible signature of synchrotron self - absorption of the strong radio outburst which occurred near epoch 17 july 2003 . at 90 cm ,
the flux density of sgr a * is @xmath1 jy , suggesting a sharp decrease in flux density at wavelengths longer than 47 cm .
the spectrum at long cm wavelengths appears to be consistent with free - free absorption by a screen of ionized gas with a cutoff @xmath2100 cm .
this cutoff wavelength appears to be three times longer than that of @xmath230 cm suggested by davies , walsh & booth based on observations in 1974 and 1975 .
our analysis suggests that the flux densities of sgr a * at wavelengths longer than 30 cm could be attenuated and modulated by stellar winds from massive stars close to sgr a*.
|
sequential decision making problems under uncertainty appear in most modern applications , such as automated experimental design , recommendation systems and optimisation .
the common structure of these applications that , at each time step @xmath0 , the decision - making agent is faced with a choice . after each decision , it obtains some problem - dependent feedback @xcite . for the so - called _ bandit _
problem , the choices are between different _ arms _ , and the feedback consists of a single scalar reward obtained by the arm at time @xmath0 . for the _ prediction _ ( or full - information ) problem
, it obtains the reward of the chosen arm , but also observes the rewards of all other choices at time @xmath0 . in both cases ,
the problem is to maximise the total reward obtained over time .
however , dealing with specific types of feedback may require specialised algorithms . in this paper
, we show that the _ thompson sampling _ algorithm can be applied successfully to a range of sequential decision problems , whose feedback structure is characterised by a graph .
our algorithm is an extension of thompson sampling , introduced in @xcite .
although easy to implement and effective in practice , it remained unpopular until relatively recently .
interest grew after empirical studies @xcite demonstrated performance exceeding state of the art .
this has prompted a surge of interest in thompson sampling , with the first theoretical results @xcite and industrial adoption @xcite appearing only recently . however , there are still only a few theoretical results and many of these are in the simplest settings . however , it is easy to implement and effective under very many different settings with complex feedback structures , and there is thus great need to extend the theoretical results to these wider settings .
@xcite argue that thompson sampling is a very effective and versatile strategy for different _
information structures_. their paper focuses on specific examples : the two extreme cases of no and full information mentioned above and the case of linear bandits and combinatorial feedback . here
we consider the case where the feedback is defined through a graph @xcite . more specifically ,
the arms ( choices ) are vertices of a ( potentially changing ) graph and when an arm is chosen , we see the reward of that arm as well as its neighbours
. on one hand , it is a clean model for theoretical and experimental analysis and on the other hand , it also corresponds to realistic settings in social networks , for example in advertisement settings ( c.f .
@xcite ) . we provide a problem - independent regret bound that is parametrized by the clique cover number of the graph and naturally generalizes the two extreme cases of zero and full information .
we present two variants of thompson sampling , that are both very easy to implement and computationally efficient .
the first is straightforward thompson sampling , and so draws an arm according to its probability of being the best , but also uses the graph feedback to update the posterior distribution .
the second one can be seen as sampling cliques in the graph according to their probability of containing the best arm , and then choosing the empirically best arm in that clique .
neither algorithm requires knowledge of the complete graph .
almost all previous algorithms require the full structure of the feedback graph in order to operate .
some require the entire graph for performing their updates only at the end of round ( e.g. @xcite ) others actually need the description of the graph at the beginning of the round to make their decision and almost none of the algorithms previously proposed in the literature is able to provide non - trivial regret guarantees without the feedback graphs being disclosed .
however , @xcite ( @xcite ) argue that the assumption that the entire observation system is revealed to the learner on each round , even if only after making her prediction , is rather unnatural . in principle
, the learner need not be even aware of the fact that there is a graph underlying the feedback model ; the feedback graph is merely a technical notion for us to specify a set of observations for each of the possible arms .
ideally , the only signal we would like the agent to receive following each round is the set of observations that corresponds to the arm she has taken on that round ( in addition to the obtained reward ) . our algorithms work in this setup
- they do not need the whole graph to be disclosed either when selecting the arm or when updating beliefs - only the local neighborhood is needed .
furthermore , the underlying graph is allowed to change arbitrarily at each step .
the detailed proofs of all our main results are available in the full version of this paper .
the stochastic k - armed bandit problem is a well known sequential decision problem involving an agent sequentially choosing among a set of k arms @xmath1 . at each round @xmath0
, the agent plays an arm @xmath2 and receives a reward @xmath3 , where @xmath4 is a random variable defined on some probability space @xmath5 and @xmath6 is a reward function .
each arm @xmath7 has mean reward @xmath8 .
our goal is to maximize its expected cumulative reward after @xmath9 rounds .
an equivalent notion is to minimize the expected regret against an oracle which knows @xmath10 .
more formally , the expected regret @xmath11 of an agent policy @xmath12 for a bandit problem @xmath10 is defined as : @xmath13 where @xmath14 is the mean of the optimal arm and @xmath15 is the policy of the agent , defining a probability distribution on the next arm @xmath16 given the history @xmath17 of previous arms and rewards .
the main challenge in this model is that the agent does not know @xmath10 , and it only observes the reward of the arm it played . as a consequence ,
the agent must trade - off exploitation ( taking the apparently best arm ) with exploration ( trying out other arms to assess their quality ) .
the bayesian setting offers a natural way to model this uncertainty , by assuming that the underlying probability law @xmath10 is in some set @xmath18 parametrised by @xmath19 , over which we define a prior probability distribution @xmath20 . in that case
, we can define the bayesian regret : @xmath21 a policy with small bayesian regret may not be uniformly good in all @xmath10 . since in the bayesian setting we frequently need to discuss posterior probabilities and expectations , we also introduce the notation @xmath22 and @xmath23 for expectations and probabilities conditioned on the current history . in this model
, we assume the existence of an undirected graph @xmath24 with vertices corresponding to arms . by taking an arm @xmath25 ,
we not only receive the reward of the arm we played , but we also observe the rewards of all neighbouring arms @xmath26 .
more precisely , at each time - step @xmath0 we observe @xmath27 for all @xmath28 , while our reward is still @xmath3 . if the graph is empty , then the setting is equivalent to the bandit problem .
if the graph is fully connected , then it is equivalent to the prediction ( i.e. full information ) problem .
however , many practical graphs , such as those derived from social networks , have an intermediate connectivity . in such cases
, the amount of information that we can obtain by picking an arm can be characterised by graph properties , such as the clique cover number : a clique covering @xmath29 of a graph @xmath30 is a partition of all its vertices into sets @xmath31 such that the sub - graph formed by each @xmath32 is a clique i.e. all vertices in @xmath32 are connected to each other in @xmath30 .
the smallest number of cliques into which the nodes of @xmath30 can be partitioned is called the clique cover number .
we denote by @xmath33 the minimum clique cover and @xmath34 its size , omitting @xmath35 when clear from the context .
the domination number is another useful similar notion for the amount of information that we can obtain .
a dominating set in a graph @xmath36 is a subset @xmath37 such that for every vertex @xmath38 , either @xmath39 or @xmath40 for some @xmath41
. the smallest size of a dominating set in @xmath30 is called the domination number of @xmath35 and denoted @xmath42 .
optimal policies for the stochastic multi - armed bandit problem were first characterised by @xcite , while index - based optimal policies for general non - parametric problems were given by @xcite .
later @xcite proved finite - time regret bounds for a number of ucb ( upper confidence bound ) index policies , while @xcite proved finite - time bounds for index policies similar to those of @xcite , with problem - dependent bounds @xmath43 .
recently , a number of policies based on sampling from the posterior distribution ( i.e. thompson sampling@xcite ) were analysed in both the frequentist @xcite and bayesian setting @xcite and shown to obtain the same order of regret bound for the stochastic case .
for the _ adversarial _ bandit problem the bounds are of order @xmath44 .
the analysis for the full information case generally results in @xmath45 bounds on the regret @xcite , i.e. with a much lower dependence on the number of arms .
intermediate cases between full information and bandit feedback can be obtained through graph feedback , introduced in @xcite , which is the focus of this paper .
in particular , @xcite and @xcite analysed graph feedback problems with stochastic and adversarial reward sequences respectively .
specifically , @xcite analysed variants of upper confidence bound policies , for which they obtained @xmath46 problem - dependent bounds . in more recent work
, @xcite also introduced algorithms for graphs where the structure is never fully revealed showing that ( unlike the bandit setting ) there is a large gap in the regret between the adversarial and stochastic cases . in particular , they show that in the adversarial setting one can not do any better than ignore all additional feedback , while they provide an action - elimination algorithm for the stochastic setting .
finally , @xcite obtain a problem - dependent bound of the form @xmath47 where @xmath48 is the linear programming relaxation to @xmath49 and @xmath50 is the minimum degree of @xmath30 .
[ [ contributions . ] ] contributions .
+ + + + + + + + + + + + + + in this paper , we provide much simpler strategies based on thompson sampling , with a matching regret bound . unlike previous work , these are also applicable to graphs whose structure is unknown or changing over time . more specifically : 1 .
we extend @xcite to graph - structured feedback , and obtain a problem - independent bound of @xmath51 .
2 . using planted partition models , we verify the bound s dependence on the clique cover .
we provide experiments on data drawn from two types of random graphs : erds rnyi graphs and power law graphs , showing that our algorithms clearly outperform ucb and its variations @xcite .
finally , we measured the performance on graphs estimated from the data used in @xcite .
once again , thompson sampling clearly outperforms ucb and its variants .
we consider two algorithms based on thompson sampling .
the first uses standard thompson sampling to select arms .
as this also reveals the rewards of neighbouring arms , the posterior is conditioned on those as well .
the second algorithm uses thompson sampling to select an arm , and then chooses the empirically best arm within that arm s clique .
the policy is an adaptation of thompson sampling for graph - structured feedback .
thompson sampling maintains a distribution over the problem parameters . at each step , it selects an arm according to the probability of its mean being the largest .
it then observes a set of rewards which it uses to update its probability distribution over the parameters . for the case where each arm has an independent parameter defining its reward distribution
, we can update the distribution of all arms observed separately .
a particularly simple case is when all the reward are generated from bernoulli distributions .
then we can simply use a beta prior for each arm , illustrated by the policy in algorithm [ alg : tsn ] .
we note that the algorithm trivially extends to other priors and families . for each arm @xmath7 , set @xmath52 and @xmath53 play arm @xmath54 @xmath55
if @xmath56 the @xmath57 , else @xmath58 the policy does not fully exploit the graphical structure .
for example , as noted by @xcite , instead of doing exploration on arm @xmath7 we could explore an apparently better neighbour , which would give us the same information .
more precisely , instead of picking arm @xmath7 , we pick the arm @xmath59 with the best empirical mean .
the intuition behind it is that , if we take any arm in @xmath60 , we are going to observe anyway the reward of @xmath7 .
so it is always better to exploit the best arm in @xmath60 .
the resulting policy , is summarized in algorithm [ alg : tsmaxn ] .
although our theoretical results do not apply to this policy , it can have better performance as it uses more information . for each arm @xmath7 , set @xmath52 and @xmath53 let @xmath61 be the empirical mean of arm @xmath7
let @xmath62 play arm @xmath63 @xmath55 if @xmath56 the @xmath57 , else @xmath58 russo and van roy introduced an elegant approach to the analysis of thompson sampling .
they define the _ information ratio _ as a key quantity for analysing information structures : @xmath64 ^ 2}{{\mathbb{i}}_t(a^ * , ( a_t , y_{t , a_t}))},\ ] ] where @xmath65 and @xmath66 denote expectation and mutual information respectively , conditioned on the history of arms and observations until time @xmath0 .
they show that it follows very generally that [ prop : ts ] if @xmath67 almost surely for all @xmath68 , then , @xmath69 here @xmath70 denotes entropy . thus to analyse the performance of thompson sampling on a specific problem ,
one may focus on bounding the information ratio ( [ eq : ir ] ) .
for the ( independent ) @xmath71-armed bandit case , they show that @xmath72 , while for full - information ( @xmath71 experts ) case , they show that @xmath73 .
we now give a simple but useful extension of their results which is intermediate between these cases .
[ prop : eq ] let @xmath74 be an equivalence relation defined on the arms with @xmath75 denoting the equivalence class of @xmath76 .
let @xmath77 for sequence of random variables @xmath78
. then @xmath79 , half the number of equivalence classes .
this is a direct generalisation of propositions 3 and 4 in @xcite , to which it reduces when the equivalence relation is trivial ( bandit case ) or full ( expert case ) .
we can now use proposition [ prop : eq ] to analyse graph structured arms : [ lem : gr ] let @xmath80 be a graph with @xmath81 corresponding to the arms and suppose that when an arm @xmath76 is played , we observe the rewards @xmath82 for all @xmath83 i.e we observe the rewards corresponding to both @xmath76 and all its neighbours .
let @xmath84 be a clique cover of @xmath30 i.e. a partition of @xmath81 into cliques .
then @xmath85 .
applying proposition [ prop : ts ] and lemma [ lem : gr ] , we get a performance guarantee for thompson sampling with graph - structured feedback : [ th : gr ] for thompson sampling with feedback from the graph @xmath30 , we have @xmath86 , where @xmath87 is the _ clique cover number _ of @xmath30 .
the bandit and expert cases are special cases corresponding to the empty graph and the complete graph respectively since @xmath88 for the empty graph and @xmath89 for the complete graph .
the _ planted partition models _ or _ stochastic block models _
graphs @xmath90 are defined as follows @xcite : first a fixed partition of the @xmath91 vertices into @xmath92 parts is chosen , then an edge between two vertices within the same class exists with probability @xmath93 and that between vertices in different classes exists with probability @xmath94 , independently with @xmath95 .
if @xmath96 , then with high probability , the clique cover number of the resulting graph is @xmath92 ( corresponding to the planted @xmath92 cliques ) .
thus for this class of graphs , the regret grows as @xmath97 as per theorem [ th : gr ] .
this is explored in section [ sec : experiments ] .
when @xmath98 but large , the planted partition graph is considered a good model of the structure of network communities .
if the underlying graph changes at each time step , then we also have the bound for the same algorithm : suppose the underlying graph at time @xmath99 is @xmath100 , then : @xmath101 the information ratio at time @xmath0 is bounded by @xmath102 .
we compared our proposed algorithms in terms of the actual expected regret against a number of other algorithms that can take advantage of the graph structure .
our comparison was performed over both synthetic graphs and networks derived from real - world data . in all our experiments
, we tested against the and algorithms , introduced in @xcite .
these are the analogues of our algorithms , using upper confidence bounds instead of thompson sampling . for the real - world networks
, we also evaluated our algorithms against a variant of @xmath103- from @xcite .
this is based on a linear program formulation for finding a lower bound @xmath104 on the size of the minimum dominating set .
we observe first that their analysis holds for _ any _ fixed dominating set @xmath105 and the bound so obtained is @xmath106 . in particular , we may use a simple greedy algorithm to compute a near optimal dominating set @xmath107 such that @xmath108 , where @xmath109 is the maximum degree of the graph .
using such a near optimal dominating set in place of the lp relaxation and choosing arms from it uniformly at random , we obtain a variant of the original algorithm , which we call @xmath103- , which is much more computationally efficient , and which enjoys a similar regret bound : the regret of @xmath103- is at most @xmath110 , where @xmath109 is the maximum degree of the graph .
@xmath103- and @xmath103- have the hyper - parameters @xmath111 and @xmath112 , which control the amount of exploration .
we found that its performance is highly sensitive to their choice . in our experiments , we find the optimal values for these parameters by performing a separate grid search for each problem , and only reporting the best results .
since there is no obvious way to tune these parameters online , this leads to a favourable bias in our results for this algorithm .- greedy performs almost always best , but its performance can degrade significantly when the parameters are changed .
although @xcite suggests a method for selecting these parameters , we find that using it leads to a near - linear regret . ]
as thompson sampling is a bayesian algorithm , we can view the prior distribution as a hyper - parameter . in our experiments
, we always set that to a beta(1,1 ) prior for all rewards . for all of our experiments
, we performed @xmath113 independent trials and reported the _ median - of - means _ estimator of the cumulative regret .
it partitions the trials into @xmath114 equal groups and return the median of the sample means of each group .
we set the number of groups to @xmath115 , so that the confidence interval holds with probability at least @xmath116 .
we also reported the deviation of each algorithm using the gini s mean difference ( gmd hereafter ) @xcite .
gmd computes the deviation as @xmath117 with @xmath118 the @xmath119-th order statistics of the sample ( that is @xmath120 ) .
as shown in @xcite the gmd provides a superior approximation of the true deviation than the standard one . to account for the fact that the cumulative regret of our algorithms might not follow a symmetric distribution ,
we computed the gmd separately for the values above and below the _ median - of - means_. in our synthetic problems , unless otherwise stated , the rewards are drawn from a bernoulli distribution whose mean is generated uniformly randomly in @xmath121 $ ] except for the optimal arm whose mean is generated randomly in @xmath122 $ ] .
the number of nodes in the graph is 500 .
we tested with a sparse graph of 2500 edges and also with a dense graph of 62625 edges .
[ [ erdsrnyi - graphs ] ] erds rnyi graphs + + + + + + + + + + + + + + + + + + in our first experiment , we generate the graph randomly using the erds rnyi model . figure [ fig : erdos : small ] and [ fig : erdos : large ] respectively show the result in the sparse and dense graph .
our first observation here is that all policies take advantage of a large number of edges as their cumulative regret is better by using the dense graph ( figure [ fig : erdos : large ] ) rather than the sparse one ( figure [ fig : erdos : small ] ) .
this confirms the theoretical result as a dense graph will have a smaller clique cover number than a sparse one .
the policy outperforms all other in both the sparse and dense graph model .
however , the performance of is very close to that of in the near complete graph . this is explained by the fact that in a near complete graph we have many cliques .
it is revealing to see that outperforms both the and policies .
[ [ power - law - graph ] ] power law graph + + + + + + + + + + + + + + + such graphs are commonly used to generate static scale - free networks @xcite . in this experiment
, we generated a non - growing random graph with expected power - law degree distribution .
show the results respectively for the dense and sparse graph figure [ fig : powerlaw : large ] and [ fig : powerlaw : small ] show the results respectively for the dense and sparse graph .
again , the policy clearly outperforms all other . in the sparse graph model ,
is beaten by at the beginning of the rounds ( @xmath123 ) , but catches and ended up beating . [
[ planted - partition - model ] ] planted partition model + + + + + + + + + + + + + + + + + + + + + + + the aim of the experiment on this model is to check the dependency on the number of cliques for each policy .
figure [ fig : plantedpartition ] shows the results where on the x - axis we have the parameter @xmath92 of the planted partition graph ( which is almost equal to the number of cliques ) on a graph with 1024 nodes . on the y - axis
we have the relative regret of each policy , i.e. the ratio between the regret of each policy with the regret of the best policy when there are two groups , for ease of comparison .
as we can see , all methods regret scales similarly .
thus , the theoretical bounds appear to hold in practice , and to be somewhat pessimistic . for a larger number of nodes
, we would expect the plots to flatten later .
our experiments on real world datasets follow the methodology described in @xcite .
we first infer a graph from data , and then define a reward function for movie recommendation from user ratings .
missing ratings are predicted using matrix factorization .
this enables us to generate rewards from the graph .
we explain the datasets , reward function and graph inference in the full version .
[ [ results ] ] results + + + + + + + figure [ fig : facebook:100 ] shows the results for the facebook graph and figure [ fig : flixster:100 ] for the flixster graph . once again , the thompson sampling strategies dominate all other strategies for the facebook and they are matched by the optimised @xmath103- policy in the flixster graph .
we notice that in this setting the gap between the ucb policies and the rest is much larger , as is the overall regret of all policies .
this can be attributed to the larger size of these graphs .
we have presented the first thompson sampling algorithms for sequential decision problems with graph feedback , where we not only observe the reward of the arm we select , but also those of the neighbouring arms in the graph .
thus , the graph feedback allows us the flexibility to model different types of feedback information , from bandit feedback to expert feedback .
since the structure of the graph need not be known in advance , our algorithms are directly applicable to problems with changing and/or unknown topology .
our analysis leverages the information - theoretic construction of @xcite , by bounding the expected information gain in terms of fundamental graph properties .
although our problem - independent bound of is not directly comparable to @xcite , we believe that a problem - independent version of the latter should be @xmath124 , in which case our results would represent an improvement of @xmath125 . in practice ,
our two variants always outperform , , which also use graph feedback but rely on upper confidence bounds .
we are also favourably compared against @xmath103- , even when we tune the parameters of the latter _
post hoc_. it would be interesting to extend our techniques to other types of feedback .
for example , the bayesian foundations of thompson sampling render our algorithms applicable to arbitrary dependencies between arms . in future work
, we will analytically and experimentally consider such problems and related applications .
finally , an open question is the existence of information - theoretic lower bounds in settings with partial feedback .
this research was partially supported by the snsf grants `` adaptive control with approximate bayesian computation and differential privacy '' ( izk0z2_167522 ) , `` swiss sense synergy '' ( crsii2_154458 ) , by the the people programme ( marie curie actions ) of the european union s seventh framework programme ( fp7/2007 - 2013 ) under rea grant agreement number 608743 , and the future of life institute .
|
we present a novel extension of thompson sampling for stochastic sequential decision problems with graph feedback , even when the graph structure itself is unknown and/or changing .
we provide theoretical guarantees on the bayesian regret of the algorithm , linking its performance to the underlying properties of the graph .
thompson sampling has the advantage of being applicable without the need to construct complicated upper confidence bounds for different problems .
we illustrate its performance through extensive experimental results on real and simulated networks with graph feedback .
more specifically , we tested our algorithms on power law , planted partitions and erds rnyi graphs , as well as on graphs derived from facebook and flixster data .
these all show that our algorithms clearly outperform related methods that employ upper confidence bounds , even if the latter use more information about the graph .
|
let @xmath7 be a fixed positive real number , and @xmath6 run over the imaginary parts of the zeros of the riemann zeta - function .
we are interested in the distribution of the fractional parts @xmath8 .
rademacher @xcite was the first to consider this problem and he conjectured that , for a certain specific type of @xmath7 , there should be a `` predominance of terms which fulfill @xmath9 '' . since the fractional parts are uniformly distributed modulo @xmath10 , as proved by hlawka @xcite in 1975 , any discrepancy must be very subtle .
the first and third authors uncovered in @xcite this delicate inequity in the fractional parts and not only proved rademacher correct , but also gave a much more precise measure for this phenomenon .
let @xmath11 be the torus .
since the fractional parts @xmath8 are uniformly distributed ( mod 1 ) for any fixed @xmath7 , for all continuous functions @xmath12 , we have @xmath13 as @xmath14 , where @xmath15 denotes the number of zeros @xmath16 . we know that (
@xcite , theorem 9.4 ) @xmath17 the first and third authors @xcite showed that , for a large class of functions @xmath12 , as @xmath14 , @xmath18 where @xmath19 is a function depending on the form of @xmath7 . from this result
, we can see that the right hand side is close to a constant , and thus the discrepancy of the set @xmath20 is of order @xmath21 . in @xcite , the first author , soundararajan , and the third author established connections between the discrepancy of this set , montgomery s pair correlation function and the distribution of primes in short intervals . in the present paper ,
our goal is to generalize the results from @xcite to the case of simultaneous distribution of fractional parts @xmath22 , where @xmath23 are distinct , positive real numbers .
as we will see below , a new phenomenon involving diophantine approximation appears in the higher dimensional case @xmath1 . for @xmath1 ,
consider a function @xmath24 defined on @xmath25 , which has fourier expansion @xmath26 and assume that @xmath27 , and @xmath28 is the sup - norm .
let @xmath29 .
suppose that there is a full - rank @xmath30 ( @xmath31 ) matrix @xmath32 with integer entries , @xmath33 for all @xmath34 , and such that @xmath35 where @xmath36 for some integers @xmath37 , @xmath38 and distinct primes @xmath39 ( @xmath40 ) . among all possible such matrices ,
we take one with maximal @xmath41 , in which case the above conditions ensure that @xmath42 is uniquely determined .
we then define @xmath43 where @xmath44 for @xmath45 .
if no such matrix exists , we simply set @xmath46 then we have the following .
[ n - dim - aa ] let @xmath1 .
assume @xmath24 satisfies condition ( [ n - dim - condt ] ) .
then , if @xmath29 satisfies @xmath47 for all @xmath48 and some fixed constant @xmath49 , we have @xmath50 * remark 1 .
* by a theorem of khintchine @xcite and theorem 2 in @xcite , the set of such @xmath51 has full lebesgue measure .
* remark 2 . *
it is possible to prove a similar conclusion with a larger class of test functions h , using more complicated fourier analysis techniques , along the lines of what was done in @xcite , or by assuming the riemann hypothesis .
for the special case when @xmath52 and @xmath53 , will always be satisfied as a consequence of baker s theorem @xcite , and we have the following . [ n - dim - as ] if @xmath24 satisfies condition ( [ n - dim - condt ] ) and @xmath51 satisfies condition ( [ n - dim - condition ] ) for @xmath52 , we have @xmath50 in general , the conclusion of theorem [ n - dim - aa ] may fail if the diophantine condition fails ; for details see the remarks at the conclusion of section [ section-2dim ] . for the case
@xmath54 , the behavior of @xmath55 is determined by the behavior of the convergents @xmath56 of the continued fraction of @xmath57 .
even if fails , which occurs when some @xmath58 are very large compared to @xmath59 , we show that there are `` long '' intervals of @xmath60-values on which we may draw the conclusion of the previous theorem .
[ 2-dim - range ] assume and fix any @xmath61 .
let @xmath62 $ ] .
then , @xmath63 finally , we show some graphs comparing the limiting density function @xmath64 with numerical computations obtained from the first 100 million zeros of the riemann zeta - function ( which were kindly provided by toms oliveira e silva , see @xcite ) . define @xmath65 take @xmath66 , then @xmath67 .
denote @xmath68 .
we partition @xmath69 into @xmath70 small rectangles with side length @xmath71 .
for some values of @xmath2 , @xmath3 , we plot the values of @xmath72 for each small rectangle @xmath73 . * example 1 .
* let @xmath2 and @xmath3 satisfy @xmath74 figure @xmath75 shows the graph of @xmath64 .
figure @xmath76 shows the values of @xmath77 , the graph of which is a close approximation of @xmath64 .
* example 2 .
* figure @xmath78 and figure @xmath79 show the analogous graphs for the case when @xmath2 and @xmath3 satisfy @xmath80 , title="fig:",width=529 ] + , title="fig:",width=529 ] + , title="fig:",width=529 ] + , title="fig:",width=529 ] +
we need lemma 1 from @xcite , which is an extension of a famous formula of landau .
let @xmath81 denote a generic nontrivial zero of the riemann zeta - function , and denote @xmath82 . by our assumption ( [ n - dim - condt ] )
, one can see that the error term above is @xmath93 .
thus , by ( [ hidim - hfunc ] ) , we have @xmath94 the second equality is a consequence of the identity @xmath95 .
let @xmath96 .
hence , by and lemma [ lem - ford2 ] , we have for @xmath60 large and @xmath97 that @xmath98 write @xmath99 applying lemma [ lem - fz ] , we get @xmath100 here we used our assumption ( [ n - dim - alpha ] ) that @xmath101 now let @xmath102 . by ( [ hidim - h ] ) , ( [ n - dim - lem ] ) and @xmath103 , we get @xmath104 since @xmath105 is bounded , the first sum is absolutely and uniformly convergent in @xmath60 under the assumption @xmath106 .
and the terms with @xmath107 tend to zero as @xmath14 .
thus , by ( [ hidim - t ] ) , as @xmath14 , @xmath108 since @xmath109 unless @xmath110 is prime power , by ( [ n - dim - g2 ] ) , when @xmath111 , the sum in the right hand side of ( [ n - dim - conclu ] ) is also @xmath112 . if @xmath51 satisfies ( [ n - dim - condition ] ) with @xmath42 being maximal ( maximal @xmath41 ) , then the only terms with @xmath113 ( that is , with @xmath110 being a prime power ) are those with @xmath114 being a multiple of some row of @xmath42
. hence , recalling the definition of @xmath64 , as @xmath115 , the right hand side of ( [ n - dim - conclu ] ) becomes @xmath116 which equals @xmath117 by ( [ n - dim - g1 ] ) .
so in all cases , we have @xmath50 for the special form of @xmath51 when @xmath52 , we can solve the above linear equations for @xmath51 , and find that each @xmath118 is a linear combination of these @xmath119 s with rational coefficients .
hence , by baker s theorem @xcite , for such @xmath51 , there exist constants @xmath120 and @xmath121 , such that , for all @xmath48 , latexmath:[\[\label{n - dim - baker } we now break the double sum into two pieces , over those pairs @xmath127 corresponding to `` large '' @xmath128 and those pairs @xmath127 corresponding to `` small '' @xmath128 .
fix an arbitrary positive constant @xmath49 and denote @xmath129 by standard facts on diophantine approximation , we know that if @xmath135 then @xmath136 and hence all elements of @xmath137 are of the form @xmath138 , where @xmath139 is a convergent of the continued fraction of @xmath140
. as @xmath59 grows at least exponentially and @xmath141 for @xmath142 , there are @xmath143 elements in @xmath137 .
let @xmath144 denote the largest @xmath145 such that @xmath142 . by elementary properties of continued fractions ( chapter x , @xcite , or corollary 1.4 , @xcite )
, we have @xmath146 thus , for all @xmath147 except possibly @xmath148 , we have @xmath149 and from lemma [ lem - fz ] we get ( writing @xmath150 ) @xmath151 where we have used that @xmath152 .
we adopt the convention that the term @xmath148 is included in the sum above if @xmath153 .
in fact , the error term from lemma [ lem - fz ] is acceptable , namely @xmath154 , if @xmath155 as well .
now let @xmath102 and assume that @xmath156 .
we revert to the sum over @xmath157 . by and
the the trivial bound for @xmath158 , we have @xmath159 combined with and , we see that if ( a ) @xmath137 is empty or if ( b ) there is some convergent of @xmath123 with @xmath160 $ ] , then @xmath161 note that ( b ) is satisfied if @xmath60 is large enough and @xmath162 $ ] .
* remarks . *
suppose that there is a very long gap between convergents , say @xmath164 is small and @xmath165 is very large
. put @xmath166 and assume that @xmath167 . using the known estimate @xmath168 , we get @xmath169 in the range @xmath170 , there is thus a term in the sum of order @xmath171 , which may be larger than @xmath60 . 99 a. baker , _ transcendental number theory _ , cambridge mathematical library ( 2nd ed . ) , cambridge university press , 1990 . y. bugeaud , _ approximation by algebraic numbers _ , cambridge university press 2004 k. ford , a. zaharescu .
on the distribution of imaginary parts of zeros of the riemann zeta function .
_ j. reine angew .
_ * 579 * , 145 - 158 ( 2005 ) k. ford , k. soundararajan , a. zaharescu . on the distribution of imaginary parts of zeros of the riemann zeta function ,
( 2009 ) * 343*:487 - 505 g. h. hardy , e. w. wright , _ an introduction to the theory of numbers , 5th ed . _ ,
oxford , 1979 e. hlawka , ber die gleichverteilung gewisser folgen , welche mit den nullstellen der zetafunktionen zusammenhngen .
wiss , _ math .- naturnw .
_ ii * 184 * , 459 - 471(1975 ) r. s. kemble , a groshev theorem for small linear forms , _ mathematika , _ * 52 * ( 2005 ) , 79 - 85 .
khintchine , einige stze ber ketternbrche , mit anwendungen auf die theorie der diophantischen approximationen , _ math .
_ , ( 1924 ) * 92 * , 115 - 125 h. rademacher , _ fourier analysis in number theory , _ symposium on harmonic analysis and related integral transforms : final technical report , cornell univ . , ithica , n.y .
( 1956 ) , 25 pages ; also in : h. rademacher , _ collected works , _ pp . 434 - 458 .
toms oliveira e silva , webpage , ` http://sweet.ua.pt/tos/zeta.html ` .
e. c. titchmarsh , _ the theory of the riemann zeta - function , 2nd ed . rev . by d.
r. heath - brown _ , clarendon press , oxford 1986 department of mathematics , university of illinois at urbana - champaign , 1409 west green street , urbana , il 61801 , usa and institute of mathematics of the romanian academy , p.o .
box 1 - 764 , bucharest ro-70700 , romania .
|
in this paper , we investigate the simultaneous distribution of the fractional parts of @xmath0 , where @xmath1 , @xmath2 , @xmath3 , @xmath4 , and @xmath5 are fixed distinct positive real numbers and @xmath6 runs over the imaginary parts of the non - trivial zeros of the riemann zeta - function .
|
the sins meeting was dominated by an interesting and often disconnected mix of observational reports on `` small scale '' structure in various interstellar environments , and theoretical works attempting to explain such observed structures .
interestingly , there was a wider variety in the kinds and physical properties of the structures reported observationally than in the theoretical approaches presented as potential models .
this article includes a review of the theoretically - significant results presented at the sins workshop .
for reference , we first recall observational results that pose the challenge for the theoretical studies .
small - scale structure was reported to be observed in all three of the diffuse ionized , atomic and molecular components of the interstellar medium ( ism ) , although the physical properties of the structures in each component are as varied as the environments in which they are found ( see the observational review by heiles & stinebring in this volume ) .
even the very notion of `` small '' differs strongly from one field to another .
the features are primarily distinguished by their observational signatures . in the diffuse ionized gas , small - scale structures
are inferred from radio scintillation effects , such as large flux density variations of compact radio sources or pronounced `` fringing '' in the dynamic spectra of pulsars .
a more subtle observational indicator is a host of effects suggesting that additional fluctuations in plasma density are needed above and beyond the kolmogorov spatial power spectrum which has become canonical .
comments addressing the existence of such excess power were made in the presentations by barney rickett and dan stinebring .
deducing physical properties of these ionized enhancements is not so straightforward , and more model - dependent than for the atomic and molecular components , but the results that are available give densities @xmath0 @xmath1 , sizes @xmath2 au , and filling factors @xmath3 ( rickett s talk ) .
the so - called extreme scattering events ( eses ) are ,
well , more extreme , with estimated densities in the range @xmath4 @xmath1 , and diameters of @xmath5 au ( clegg , fey , and lazio 1998 ) .
the structures responsible for ese could be more extreme versions of the structures responsible for the ubiquitous pulsar scintillation arcs , which also seem to involve discrete , identifiable clouds rather than a continuum of turbulent fluctuations ( stinebring s talk ) .
most of the presentations dealt with observations of measurable changes in the neutral hydrogen column density over small angular distances on the sky .
interpreted at face value , these features would correspond to small ( @xmath6 10 - 100 au in size ) regions of greatly enhanced gas density ( see crystal brogan s talk ) .
characteristic values of the gas density would be @xmath7 @xmath1 as opposed to @xmath8 for the typical warm neutral medium ( wnm ) , or @xmath9 for the cold neutral medium ( cnm ) . even with the assumption of low neutral hydrogen temperatures of @xmath10 k , these regions
would then exceed the typical pressure in the neutral medium by @xmath11 orders of magnitude .
the filling factor of tsas is , however , very low , @xmath12 .
while attention at this workshop concentrated on the cnm , wnm , and dig ( diffuse ionized gas ) components of the interstellar medium , small - scale molecular structures were also reported at the conference by andreas heithausen , with typical densities @xmath13-@xmath14 , temperatures @xmath0 - 20 k , sizes @xmath15 - 5000 au , and masses @xmath3 - 1 @xmath16 .
such masses are much smaller than their virial masses ( @xmath17 ) , implying that these structures can not be gravitationally bound , and must be formed or confined by some form of pressure ( possibly ram pressure ) .
edith falgarone and pierre hily - blant further pointed to tiny molecular clouds in which the hco@xmath18 abundances are much higher than those expected in steady - state models and in which abundance ratios vary wildly , suggesting a very dynamical state .
edith interpreted these data in terms of turbulence - induced chemical reactions leading to hco@xmath18 , resulting from localized heating caused by turbulent dissipation .
the main theoretical challenge in understanding these structures is that , if they are true density enhancements , then they are strongly overpressured with respect to the ambient ism pressure , and , naively , they should either not be there , or disperse quickly .
possible explanations presented at the meeting included persistent turbulent production of transient structures , geometrical effects , or statistical observational projection effects . in
[ sec : theor_interp ] we summarize these possibilities . if the structures are really small - scale density and pressure enhancements in the ism , then the challenge is to understand how one can generate such large excursions out of a medium with typical densities between @xmath19 ( the diffuse ionized gas or dig ) and @xmath20 @xmath1 ( the wnm ) . for
the scintillation features in ionized gas , the fact that this gas is ionized removes any possibility of the pressure being reduced by a low gas temperature .
since the warm ism is characterized by a transonic velocity dispersion , theoretical explanations for the sins as actual density enhancements rely on transonic turbulence in a cooling , magnetized medium .
there are basic questions about how one theoretically describes the gas that comprises the interstellar medium , and different descriptions may be necessary for the different phases .
ellen zweibel discussed the large variety of spatial scales that arise in the gas due to viscosity , resistivity , and collisions between ions and neutrals .
in addition , there are still remaining basic physics issues about how one describes the true viscosity in a magnetized , collisionless plasma .
the ese structures occur on ( very roughly ) the ion - neutral collisional scale in the dig , i.e. the scale that corresponds to the wavelength of an alfvn wave with a frequency equal to the ion - neutral collision frequency .
perhaps such fundamental plasma scales define the sizes of tsas and ese phenomena .
the study of turbulence in plasmas is also highly relevant to the small structures in the interstellar medium , since the phenomena we observe probably grow out of the general turbulence field .
if this is the case , we need to fully understand some of the topics discussed by alex lazarian , stanislav boldyrev and joanne mason , such as the location and extent of the inertial subrange and dissipation range of the turbulence , and the relationship between different turbulence variables , like density and magnetic field .
the interstellar medium may provide a new arena for the study of plasma turbulence .
much of plasma turbulence theory has depended on spacecraft observations of the solar wind for data support . while this theory - observation link has proven fruitful
, other turbulent media in astrophysics might differ in important ways .
solar wind turbulence at 1 au is dynamically young , nearly collisionless , and describable by a simple energy equation .
the interstellar medium is different in each of these categories , so the lessons learned from studies of the solar wind might not be applicable here .
another specific difference , which might illuminate fundamental differences , is in the presence or absence of compressibility . in solar wind turbulence ,
the density fluctuations are small relative to those of magnetic field and flow velocity .
the ism , on the other hand , is characterized by a range of densities and temperatures spanning many orders of magnitude , and its cooling properties render it highly compressible and even thermally unstable in the range @xmath21 .
a number of investigators presented numerical simulations of dynamical compressions at moderate mach numbers ( @xmath22 ) ( enrique vazquez - semadeni , patrick hennebelle ) or global transonic turbulence in the wnm ( adriana gazol ) , subject to thermal bistability ( e.g. , * ? ? ? * ) , showing that in principle a population of small - scale , high - density , high - pressure objects can be transiently formed in such flows .
the basic principle is that the transonic compressions in the wnm nonlinearly trigger thermal instability , causing a phase transition to a cold , dense phase @xcite .
since the total ( thermal + ram ) pressure is higher than the thermal pressure of the wnm ( @xmath23 ) alone , the cold cloudlets end up at pressures higher than @xmath23 as well @xcite .
moreover , the compressed layers develop turbulence ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and , since they are much colder than the wnm ( @xmath24 k ) , this turbulence is strongly supersonic , so that the flow there is highly compressible and the highest - density structures are transient . in the simulations , pressures up to @xmath25 k@xmath26 and densities up to @xmath27
are readily reached ( see adriana s and patrick s contributions ) .
it is noteworthy that structures of a given high density ( e.g. , @xmath28 ) can span roughly two orders of magnitude in thermal pressure ( @xmath29-@xmath30 k@xmath26 ) , indicating that dense cloudlets can exist at both high and low temperatures . presumably , the warmer ones are those that have not had time to cool down yet , and are probably in the process of doing so .
for example , a cloudlet at @xmath31 and @xmath32 k@xmath26 is at a temperature @xmath33 k and cooling strongly .
it is already much cooler than the wnm , but not yet at the temperatures characteristic of the cnm or the molecular gas .
this suggests that indeed these structures are transient , in particular the warmer ones . in principle , then , dense , overpressured , small - scale structures ( both cold and warm ) can be generated in such flows .
patrick reported densities up to @xmath34 and pressures up to @xmath35 k@xmath26 in 2d simulations with a resolution of 10000 grid points per dimension , down to scales of hundreds of au .
this appears very close to explaining the physical conditions of _ individual _ structures reported by the atomic and molecular observations .
however , in order to determine whether the mechanism can account for the _ entire population _ of structures , detailed statistical comparisons are necessary .
adriana reported 7% of the mass being at @xmath36 , although patrick warned that this fraction seems to depend on resolution .
in general , it is necessary to measure this fraction , as well as the filling factors , as a function of the threshold density used to define the clumps in the simulations in order to compare in detail with the observations .
this should allow a quantitative determination of whether the mechanism of transient , out - of - equilibrium density fluctuations by transonic turbulence in the wnm can account for the _ population statistics _ of these objects , and not only the existence of a few such objects in isolation .
other means of comparison based on the energy and density spectra as derived taking into account projection effects were outlined by alex lazarian .
if comparisons prove to be statistically correct , then the mechanisms can be considered viable in principle , although much higher resolution and the inclusion of the relevant microphysical processes will still be needed for the simulations to reach the scales of the ionized structures responsible for radio scintillation . on the other hand ,
if the simulations do not produce these structures in sufficient numbers , then it may be necessary to explain the remainder of the observed structures by other mechanisms , such as those discussed below in [ sec : geom ] and [ sec : projec ] much of the fundamental theoretical description of the interstellar medium has been based on the concept of phases arising from thermodynamic equilibrium .
the identification of the cold and warm phases of the interstellar medium with portions of the equilibrium @xmath37 relationship dates back nearly 40 years @xcite and is appealing because of its physical simplicity .
however , there is presently a controversy , expressed indirectly at this meeting by the ( sometimes loud ! ) discussions between patrick hennebelle and enrique vzquez - semadeni , as to whether these concepts are meaningful , or if the real interstellar medium is always so far from thermodynamic equilibrium as to make those concepts invalid .
the physics of the actual ism is probably a mixture of these concepts . in regions of strong turbulent mixing ,
the interfaces between the equilibrium phases may be blurred , the flow being more similar to a density continuum than to a two - phase medium .
conversely , in more quiescent regions , thermal instability may be free to act and generate two - phase structure .
indeed , the simulations show both kinds of situations , with cnm cloudlets sometimes surviving for long times relatively unperturbed , and some others rapidly dispersing away back into the diffuse gas . in regions where two - phase structure is present ,
studies of clump evaporation are relevant , and were presented by jonathan slavin , inoue san and nagashima san , with reports of typical evaporation timescales @xmath20 myr . however , since the small - scale structures that are the subject of this conference are generally agreed to be strongly overpressured , it is expected that their lifetimes are determined by a dynamical formation and re - expansion process , with the characteristic timescale being their turbulent crossing time , which is generally much shorter than the evaporation time . in regions where two - phase structure is present , the development of thermal instability may also generate moderate turbulence ( inutsuka & koyama s poster ) .
one disappointment in the workshop was the lack of a more vigorous and extensive discussion about the relationship ( if any ) between the atomic and ionized entities .
barney rickett and mark walker are to be complimented as progressives in this respect . in coming to an eventual understanding of these objects
, it will be helpful to know if the radio propagation effects are caused by the ionized outer envelopes of the clouds responsible for tsas , or if we are dealing with entirely distinct structures that reside within the cnm or wnm on one hand , and the dig on the other .
the proper primitive equations for study of the interstellar medium are the equations of magnetohydrodynamics , not just hydrodynamics .
the interstellar medium is permeated by a magnetic field with a magnitude of 3 - 5 @xmath38 g , resulting in a plasma @xmath39 ( defined as the square of the ion - acoustic to alfvn speed ratio ) of less than 1 .
the magnetic field pressure and tension play a dynamic role in the interstellar medium , and can not be ignored .
several presentations illustrated the role played by interstellar magnetism .
alex lazarian discussed some intrinsically magnetic effects , such as the generation of density fluctuations by mhd waves even at scales smaller than the viscous scales .
adriana gazol presented results from simulations of thermally bistable , magnetized turbulent flows indicating that in this case a population of cold regions of low density ( i.e. , very low thermal pressure ) supported by magnetic pressure exists in the flow .
the question then comes down to how much variation in @xmath40 can be mechanically balanced by @xmath41 .
the value of @xmath42 for @xmath43 is @xmath44 dynes/@xmath45 , corresponding to an @xmath46 product of @xmath47 k/@xmath48 .
large variations in the magnetic energy density could therefore balance variations of several thousand ( cgs units ) in the dimensional quantity @xmath46 .
it is widely believed that , due to flux freezing , the magnetic field should be enhanced together with the density fluctuations .
however , both observations ( e.g. , * ? ? ? * ; * ? ? ?
* crystal brogan s talk at this meeting ) and numerical simulations of mhd turbulence ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) show otherwise .
the magnetic field strength appears to be in general uncorrelated with the density , except perhaps at the highest densities . in his talk ,
enrique vzquez - semadeni recalled a proposed explanation for this phenomenon @xcite based on the fact that the different modes of _ nonlinear _ mhd waves ( so called `` simple '' waves ) are characterized by different scalings of the field strength with density , so that in a turbulent medium the field strength at a given point does not directly depend on the density , but on the ( random ) history of wave passages through that point .
it had been previously shown already @xcite that compressions of a thermally bistable medium oblique to the magnetic field up to a certain angle ( that depends on the mach number of the compression ) can induce a transition to the dense phase without increasing the field , because the latter re - orients the motions along itself , and the matter can slide freely along the field .
thus , the lack of correlation is to be expected even in ideal mhd situations .
in addition , and in some sense prior to this , we must be certain that the observations really indicate relatively isotropic regions of enhanced density , rather than highly anisotropic regions such as sheets , which are fortuitously aligned with the line of sight .
this possibility was first discussed by carl heiles in 1997 . in this case
, the implied volume densities and pressures would not be as extreme as the face - value implications of the observations .
statistical analyses are again necessary to determine whether the chance alignment of the structures is sufficient to account for the observed frequency of the phenomenon .
a. deshpande proposed an explanation for tsas and ese that contrasts greatly with all of the other presentations , which attribute these phenomena to compact , overdense gaseous structures in the interstellar medium .
desh contends that the observations of large differences in the optical depth on two lines of sight separated by a distance @xmath49 are not due to `` clouds '' with physical sizes @xmath49 , but rather to the statistics of the optical depth structure function in a medium with a spatial power law in neutral hydrogen density .
desh emphasized that the optical depth structure function at a spatial lag @xmath49 ( essentially what is being measured ) is not determined solely by structures with size @xmath49 , but in general contains contributions from density structures on all scales .
desh s presentation drew largely on previous results @xcite for the tsas discussion .
he also claimed that a similar reasoning would apply to path integrals through the ionized ism which produces ese and other strong scintillation phenomena .
this last contention is not obvious because optical depth is directly describable as a stochastic integral , but scintillation phenomena must be described by wave propagation through a random medium . desh s proposal is controversial because it differs so substantially from the commonly - held view that tsas , ese , and related phenomena are due to discrete and extraordinary structures in the interstellar medium .
in contrast , desh suggests that the observations are essentially statistical fluctuations resulting from a medium with a wide range of spatial scales .
if he is correct , the highly dense , overpressured objects which attracted so much discussion in this meeting would be nonexistent , or at least possess much less extreme properties .
this disturbing prospect should motivate the workshop participants ( and others ) to explore the mathematics of optical depth differences in a turbulent medium , and scrutinize their data for independent evidence of high pressures and densities in the ism , or alternatively , indications that such extremes are absent .
one of the useful features of a workshop like this is the guidance it can provide for future research investigations , the construction of new instrumentation , and the development of new computer codes .
the following are the issues which we think should attract the attention of the community in the next five years or so . *
as discussed by dave meyer , the upgrade to the hst should permit measurements of column density variations to more , and more closely - spaced stars , including binaries and globular clusters that have not been observed until now .
this development will give us our best chance to visualize the features responsible for spatially - variable absorption .
* future developments in radio astronomy would help in improving our understanding of both the tsas and ese phenomena .
the expanded very large array ( evla ) in the short term , and the square kilometer array ( ska ) in the long term will allow measurements of hydrogen absorption along more lines of sight to extragalactic radio sources and pulsars . *
the presentations and discussions at this meeting indicated that simulators should try and quantify the population statistics of the small - scale density structures in the simulations .
they should analyse their simulations so as to produce numbers on filling factors and mass fractions as a function of density threshold .
extraction of information on clump lifetimes would also be of interest .
all of this information can help us decide if highly transient , overpressured `` clouds '' in the simulations are compatible with the statistics on lines of sight occupied by tsas and ese . *
a turbulent plasma is characterized by more than density and its variations .
the more interesting fluid variables are flow velocity , magnetic field , vorticity , and the like .
optical and radio observers should be encouraged to `` get sophisticated '' and think about ways in which observations might yield new fluid variables such as vorticity . *
a related appeal ( which was brought up in the group discussion at the end of the meeting ) is for simulators to develop diagnostics which would permit a more direct comparison with observations .
one way would be to calculate path integrals through their simulations , or at a more advanced level , spectral line profiles
. such diagnostics would facilitate the conversation between observers and theorists , and provide information the observers need .
as jim cordes said , speaking of the relationship between the scattering measure ( defined as the path integral of the turbulence parameter @xmath50 ( proportional to the variance of the density fluctuations ) ) and the emission measure , `` if you have emission measure , you can turn off the scattering measure , but you ca nt have a scattering measure and turn off the emission measure '' .
the authors are glad to acknowledge useful comments from adriana gazol .
all the conference participants appreciated and will long remember the excellent organization and arrangements made by nrao , as well as the splendid environment of socorro and central new mexico .
we particularly want to recognize the fine work of terry romero and miller goss .
evs thanks conacyt for support under grant 47366-f .
srs thanks the us national science foundation for support under grant atm03 - 54782 .
|
the workshop on small ionized and neutral structures in the interstellar medium featured many contributions on the theory of the objects which are responsible for `` tiny scale atomic structures '' ( tsas ) and `` extreme scattering events '' ( ese ) .
the main demand on theory is accounting for objects that have the high densities and small sizes apparently required by the observations , but also persist over a sufficiently long time to be observable .
one extensively - discussed mechanism is compressions by transonic turbulence in the warm interstellar medium , followed by thermal instabilities leading to an even more compressed state . in addressing the requirements for overpressured but persistent objects
, workshop participants also discussed fundamental topics in the physics of the interstellar medium , such as the timescale for evaporation of cool dense clouds , the relevance of thermodynamically - defined phases of the ism , the effect of magnetic fields , statistical effects , and the length and time scales introduced by interstellar processes .
|
v1141 aql was suspected for variability by kukarkin
_ et al . _ ( 1951 ) and confirmed as a variable star by kukarkin _
et al . _ ( 1968 ) .
it was later classified as a dwarf nova by vogt and bateson ( 1982 ) . in subsequent documentation ,
v1141 aql is only referred to as a dwarf nova ( bruch _ et al .
_ 1992 , downes _
1997 ) .
we learnt about the continuing outburst of v1141 aql from the vsnet e - mail http://vsnet.kusastro.kyoto-u.ac.jp/vsnet/mail/vsnet-outburst/msg04198.html ] reporting the observation of maciej reszelski and indicating that the star reached a magnitude of 15.2 at 2002 july 8.927 ut .
previous brightenings of v1141 aql were reported on 2001 july and 2000 september ( reszelski 2002 , private communication ) .
observations of v1141 aql reported in the present paper were obtained over five nights during the period from 2002 july 9 to 14 at the ostrowik station of the warsaw university observatory . the 60-cm cassegrain telescope equipped with a tektronics tk512cb back illuminated ccd camera was used .
the scale of the camera was 0.76" / pixel providing a @xmath2 field of view .
the full description of the telescope and camera was given by udalski and pych ( 1992 ) .
we monitored the star in `` white light '' .
this was due to the lack of an autoguiding system , not yet implemented after recent telescope renovation .
thus we did not use any filter because we wanted to have the shortest possible exposure times and avoid elongation of the star shapes .
of course , we can not include v1141 aql into the toads because according to howell _ et al . _
( 1995 ) the main property of this group is the amplitude of the superoutbursts higher than 6 mag . in the case of v1141 aql
this amplitude is around 5 mag , suggesting that this star may be a link between ordinary su uma stars and toads .
in addition to the clearly visible modulation connected with the superhumps we have also detected the weak signal characterized by an amplitude around 0.01 mag and a period of 0.03923(8 ) days ( @xmath3 min ) .
the nature of this periodicity is unknown .
this can not be the orbital period of the binary system because , according to the schoembs & stolz ( 1984 ) relation , it should be around 84 min .
it could be a signature of the white dwarf rotation period , which in nonmagnetized systems such as su uma stars , is not synchronized with the orbital period . ,
2000 , _ pasj _ , * 52 * , 429 , 1992 , _ aas _ , * 93 * , 463 , 1997 , _ pasp _ , * 109 * , 345 , 1995 , _ apj _ , * 439 * , 337 , 1992 , _ pasj _ , * 44 * , l215 , 1998 , in _ 13th north american workshop on cataclysmic variables _ , eds .
howell , e. kuulkers , c. woodward , asp , san francisco , 34 , 1951 , _ catalogue of suspected variable stars _ , , moscow , acad . of sciences
ussr shternberg , 2001a , _ pasj _ , * 53 * , 1191 , 2001b , _ pasj _ , * 53 * , 893 , 1968 , _ ibvs _ , * 311 * , 1 , 1997 , _ apj _ , * 490 * , 840 , 1998 , _ pasj _ , * 50 * , 297 , 1997 , _ acta astron .
_ , * 47 * , 281 , 1977 , _ apj _ , * 214 * , 144 , 1987 , _ apj _ , * 313 * , 772 , 1996 , _ apj letters _ , * 460 * , l107 , 1997 , _ acta astron . _ ,
* 47 * , 201 , 1987 , _ pasp _ , * 99 * , 191 , 1984 , _ astron . &
astrophys . _ ,
* 132 * , 187 , 1992 , _ acta astron . _ , * 42 * , 285 , 1982 , _ aas _ , * 48 * , 383 , 1978 , _ mnras _ , * 182 * , 777
|
the results of the ccd observations of the 2002 superoutburst of v1141 aql are described .
we have detected clear superhumps characterized by the period of 0.05930(5 ) days .
there was another , much weaker , modulation in the light curve of v1141 aql .
its amplitude is equal to @xmath0 mag and its period to 0.03923(8 ) days .
we have also discovered quasi periodic oscillations ( qpos ) with a mean period equal to 130 s and amplitude of @xmath1 mag .
* key words : * stars : individual : v1141 aql binaries : close novae , cataclysmic variables
|
oh megamasers ( ohms ) trace of some of the most extreme physical conditions in the universe - in particular , the presence of an ohm signals specific stages in the merger process of gas - rich galaxies .
ohms can thus be used as probes of their environments , both directly and indirectly .
characteristics of the maser emission itself can be used to measure extragalactic magnetic fields ( via zeeman splitting ) and gas kinematics , while the _ presence _ of an ohm is a signpost for phenomena associated with galaxy mergers , including extreme star formation and merging black holes .
ohms are a unique tool in this respect due to their extreme luminosities and ability to be seen at cosmic distances .
the total number of ohms detected to date is still low . as of 2012 , there are @xmath9 ohms published in the literature , with roughly 50% discovered in the arecibo survey of ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* darling & giovanelli ( 2000 , 2001 , 2002a ) ) .
no ohms at a distance of greater than 1300 mpc ( @xmath10 ) have been detected , and no large , systematic searches for high-@xmath11 ohms have been carried out .
the association of ohms with ir - bright merging galaxies , however , means that the density of ohms is expected to be much higher at @xmath12 , coinciding with an increase in both merging rate and cosmic star formation .
we have conducted a search for high - redshift ohms using the green bank telescope ( gbt ) .
we constructed three samples of galaxies to search for high - redshift ohms .
none of the samples are fully complete or flux - limited , but draw on the catalogs of ir - luminous galaxies with well - defined redshifts that were available at the time .
the first sample of galaxies consisted of iras sources included in the redshift catalog of the pscz survey .
galaxies were selected according to similar criteria as in the flux - limited arecibo sample , but included objects lying outside the declination limits of arecibo ( @xmath13 ) .
galaxies selected for gbt observations had : a declination range of @xmath14 or @xmath15 , a redshift range of @xmath16 , and a lower - luminosity threshold of @xmath17 .
153 galaxies in the pscz met these criteria , of which 47 of the brightest candidates were observed according to the lst windows during the early commissioning phase of the gbt in 2002 . the second sample of potential ohm hosts was assembled from flux - limited catalogs of ulirgs at higher redshifts .
we began with 35 galaxies in the fsc - first catalog @xcite , which consists of targets detected in both the iras faint source catalog and the 20-cm vla first survey .
this was supplemented with 5 ir - bright galaxy pairs and 26 sub - millimetre galaxies .
all galaxies have @xmath18 , with more than half having @xmath19 .
the highest redshift in this sample is at @xmath20 .
the third group of 19 ohm candidates was the last observed in our program , and made explicit use of the results from the first two samples .
targets were selected from the cosmos field , a 2-deg@xmath21 survey with deep spectral coverage from x - ray through radio wavelengths .
recent infrared @xcite studies show that while ohms are found in infrared - bright galaxies , the ohm fraction is much higher for starburst - dominated galaxies vs. agn .
selection of ohm candidates began with cosmos galaxies detected by _ spitzer _ at 70 @xmath22 m , and then eliminating all targets except the lirgs , ulirgs , and hylirgs identified by @xcite .
we removed all galaxies identified as agn or with @xmath23 .
finally , we culled the target list based on the expected rfi conditions near the observed frequency bands . by limiting the observed oh frequencies to cleaner regions
, we have a broader margin for error on the galaxy redshift .
the two windows used are at @xmath24 mhz and @xmath25 mhz , equivalent to redshifted oh at @xmath26 and @xmath27 .
* observations : * we observed the ohm candidates in several sessions at the gbt from 20022010 , totaling approximately 150 hours .
the majority of observations used the maximum available bandwidth of 50 mhz and 8192 channels .
integration times were selected with the goal of achieving @xmath28 mjy rms per channel for each galaxy .
the data were reduced using standard routines in gbtidl , including extensive flagging for rfi . after flagging
, we fit the radio continuum around the expected line center with a polynomial function of order @xmath29 .
this removed both intrinsic continuum structure from the target itself and any baseline structure not removed by the position - switching technique .
after stacking , the spectra were smoothed to a rest - frame velocity resolution of 10 km s@xmath30 .
.properties of new ohm detections from the gbt survey [ cols="<,<,>,^,^,^ " , ]
@xmath31 km s@xmath30 ohm at a range of telescope rms sensitivities as a function of redshift .
, width=518 ] out of 128 galaxies observed for oh , we detected new oh megamaser emission in nine objects ( figure [ fig1 ] ) . seven of the detections were pscz galaxies from the first sample , while the other two were from the fsc - first catalog in the second sample .
all nine ohms have redshifts near the lower end of the sample distribution , with the most distant lying at a redshift of @xmath32 .
the observed frequencies are in the range of @xmath33 mhz , which is covered by the l - band receiver and has relatively little rfi compared to the gbt prime focus bands .
the oh emission has been confirmed with gbt follow - up observations for five of the galaxies .
we also confirmed the detection of the previously - discovered ohm iras 09539 + 0857 @xcite .
table [ tab1 ] lists the 18-cm radio properties of the new ohm detections .
we give the galaxy s optical redshift ( @xmath34 ) , peak flux density of the ohm ( @xmath35 ) , the ratio of the integrated 1667 mhz emission to its peak flux density ( @xmath36 ) , measured oh luminosity ( log @xmath37 ) , and predicted oh luminosity ( @xmath38 ) based on the @xmath39 relationship in @xcite .
112 galaxies showed no confirmed detections of oh . the @xmath40 for each galaxy was ( conservatively ) derived assuming a boxcar line profile with a linewidth @xmath41 km s@xmath30 and a 1.5@xmath42 detection .
the rms was measured from baseline - subtracted continuum centered on the optical redshift of the galaxy and in a frequency range sufficient to cover the uncertainty in the optical redshift ( @xmath43 $ ] ) .
one of the goals of performing a search for ohms at higher redshifts was to improve the measurements of the oh megamaser luminosity function ( lf ) .
@xcite used the results of the flux - limited arecibo survey to construct a well - sampled lf between @xmath44 , which followed a power law in integrated line luminosity of @xmath45\propto~l_{oh}^{-0.64}$ ] mpc@xmath46 dex@xmath30 .
this measurement was limited to a narrow redshift range , spanning @xmath47 .
we constructed a new oh lf by combining the gbt and arecibo ohm detections , using the @xmath48 method and combining limits on both spectral line and continuum emission . we fit a power - law to all bins with more than one detection , yielding : @xmath49=(-0.66\pm0.14)~{\rm log}~l_{oh } - ( 4.91\pm0.41),\ ] ] where @xmath50 is measured in mpc@xmath46 dex@xmath30 and @xmath37 in @xmath51 .
the new detections only change the slope measured by @xcite by @xmath52 and the offset by @xmath53 .
both values are well within the uncertainties of the combined lf , as well as that of the original arecibo lf . assuming the malmquist - corrected relationship of @xmath54 from @xcite , this gives @xmath55\propto l_{ir}^{-0.83\pm0.18}$ ] .
we compare this to the lf of ulirgs in the local universe using the akari measurements of @xcite .
folding in the ohm fraction derived from the combined arecibo and gbt samples results in a slope of @xmath56 , a much shallower value than that measured from the akari galaxies @xmath57 .
this inconsistency may suggest either that ohms are highly saturated or that the maser strength is only weakly correlated with global properties such as @xmath58 ( * ? ? ? * ( darling & giovanelli 2002a ) ) .
the ohm lf of @xcite assumed a quadratic oh - ir relation corresponding to unsaturated masing ; a decrease in saturation could potentially steepen the ohm lf up to a slope of @xmath59 .
one of the ultimate goals of high - redshift ohm surveys is to use megamasers as tracers of the populations of merging galaxies as a function of redshift .
models of the merger rate as a function of redshift are typically parameterized with an evolutionary factor of @xmath7 .
the value of @xmath60 , however , is not well - constrained ( * ? ? ? * ; * ? ? ? * ( _ e.g. , _ kim & sanders 1998 ; bridge 2010 ) ) . sufficiently deep surveys of ohms can provide an independent measurement of the parametrization of the merging rate .
we calculated the predicted sky density of ohms as a function of redshift : @xmath61 = \frac{c d_l^2}{h_0\nu_0\sqrt{(1+z)^3\omega_m + \omega_\lambda } } \left(\frac{b}{a~{\rm ln } 10}\right ) \times \nonumber \\ \left((l_{oh , max})^a - ( l_{oh , min})^a\right),\ ] ] where @xmath62 and @xmath63 are parameters of the ohm lf from equation [ eqn - ohlf_combined ] ( @xmath64=b l_{oh}^a$ ] ) , @xmath65 is the minimum oh luminosity that can be observed at a given sensitivity level , and @xmath66 is the upper physical limit on ohm luminosity ( * ? ? ? * ( darling & giovanelli 2002b ) ) .
using the upper limit of zero ohms detected in the cosmos field , we place an upper limit on the merger rate of @xmath67 .
while this is still within uncertainties for the highest estimated values of @xmath60 , the cosmos limit is an important first step in using ohms as an independent tracer .
further measurements of oh deep fields at higher redshifts will be crucial for a more accurate constraint .
the final results for the gbt ohm survey yielded a much lower detection rate ( 9/121 = 7% ) of megamasers than expected . based on the high detection fraction for galaxies with @xmath5 from the arecibo survey ,
in addition to our careful selection of starburst - dominated galaxies , we had predicted a success rate of 2030% .
we attribute one of the primary causes of the low ohm fraction to be the sensitivity of the observations .
figure [ fig1 ] shows the upper limits for ohm candidates from our survey , along with the necessary rms sensitivity to detect oh lines as a function of redshift . to detect median - luminosity ohms at @xmath68
will require rms levels of 100 @xmath22jy , and perhaps significantly more time devoted to individual targets .
it must also be mentioned that rfi is a major culprit at @xmath69 ghz , significantly restricting the redshift path and increasing the noise in each receiver band .
future ohm searches may benefit from interferometric observations ( for which celestial rfi is uncorrelated from dish to dish ) , or from observations in more radio - quiet environments .
the gmrt ( india ) and askap ( australia ) are promising instruments for interferometry , while the upcoming 64-m sardinia radio telescope ( italy ) and 500-m fast ( china ) will be options to continue the search for ohms at high redshifts .
this work was part of the ph.d .
thesis of kww at the university of colorado , and will be submitted in early 2012 as willett , darling , kent , & braatz ( 2012 ) .
observations were funded in part by student support grants from nrao .
we are indebted to the staff of the green bank telescope , whose expertise and assistance made these observations possible .
the national radio astronomy observatory is a facility of the national science foundation operated under cooperative agreement by associated universities , inc .
|
we have carried out a search for 18-cm oh megamaser ( ohm ) emission with the green bank telescope .
the targeted galaxies comprise a sample of 121 ulirgs at @xmath0 , making this the first large , systematic search for ohms at @xmath1 .
nine new detections of ohms are reported , all at redshifts @xmath2 . for the remainder of the galaxies ,
observations constrain the upper limit on oh emission ; this rules out ohms of moderate brightness ( @xmath3 ) for 26% of the sample , and extremely bright ohm emission ( @xmath4 ) for 73% of the sample .
losses from rfi result in the ohm detection fraction being significantly lower than expected for galaxies with @xmath5 .
the new ohm detections are used to calculate an updated oh luminosity function , with @xmath6 ; this slope is in agreement with previous results .
non - detections of ohms in the cosmos field constrain the predicted sky density of ohms ; the results are consistent with a galaxy merger rate evolving as @xmath7 , where @xmath8 .
|
it is well known that the vast majority of stars belong to a binary or a multiple system , irrespective of their spectral type .
recent surveys with improved astrometric accuracy , either from space or from the ground , present clear evidence that the higher the accuracy , the larger the number of binary detections .
up to 3000 new binaries have thus been discovered during the hipparcos satellite mission ( lindegren 1997 ) .
the frequency of binaries is estimated to be at least 60% in the solar neighbourhood ( duquennoy & mayor 1991 ) but this is probably an underestimation as modelisation tends to show .
for example , odenkirchen and brosche ( 1999 ) found that a frequency of at least 70% was needed in some models to account for the existence of another 2400 ( up to now undetected ) astrometric binaries in the hipparcos catalogue .
the same is also true with respect to the improvement of the photometric accuracy : the higher the accuracy , the larger the number of variable star detections .
the results of the hipparcos photometric survey are again quite illustrative : some 8000 new variable stars have been identified ( van leeuwen 1997 ) . it is thus not at all a surprise to encounter pulsating @xmath0 scuti stars as components of stellar systems .
the catalogue of seeds and yanchak ( 1972 ) contains a third of double stars over a total of 155 @xmath0 sct variables or suspected ones .
szatmry ( 1990 ) also quotes a binary fraction of about 30 % . to study such cases
presents some obvious advantages : we have additional constraints on some important physical parameters such as distance , mass , radius ... but this of course will depend on the type of binary or multiple system of which the pulsating star is a member .
we may find them among the wide visual binaries ( vb ) , the closer visual binaries with orbital motion ( vb / o ) , the ( unresolved ) astrometric binaries ( ab ) , the spectroscopic binaries ( sb ) , or even eclipsing and ellipsoidal binary systems ( e ) . in almost all these cases
it is generally accepted that both components pertaining to the system originate from the same parent cloud , therefore have the same age and chemical composition .
additional information on the non - variable component can then be extrapolated unto the variable one and used to better constrain the variability characteristics such as the pulsation type and modes .
but we could go further and address such questions as : does binarity modify the pulsation properties or even trigger the pulsation ?
is there any correlation between rotation , pulsation and orbital motion ?
+ for all these reasons the identification of such objects offers an interesting challenge toward a better comprehension of pulsation in the @xmath0 scuti instability strip . a comparative study between the two companions allows to restrict for example the domain in mass in which the @xmath0 sct phenomenon takes place .
various recent studies exist that investigate the pulsating @xmath0 scuti stars in open clusters for similar reasons ( e.g. the praesepe cluster ( pea et al .
1998 ; alvarez et al . 1998 ) ) .
one should however also keep in mind that star formation in some open clusters is not always a single - epoch process ( e.g. m16 ( hillenbrand et al . 1993 ) ) .
we here summarize the available information for as complete a sample as possible of @xmath0 scuti stars in stellar systems in two tables , one for the binarity and one for the variability characteristics .
the basis was drawn on the catalogue work by garca et al .
( 1995 ) , supplemented by our own list and literature search .
the catalogue of the components of double and multiple stars ( ccdm , dommanget & nys 1995 ) was used to retrieve information on the visual components of the stellar systems .
table 1 . lists the name and the hd number ( or bd number if hd does not exist ) , the hipparcos number , the v magnitude , remarks concerning the multiplicity , the hipparcos parallax and error .
next comes the information regarding the duplicity in the hipparcos catalogue ( esa 1997 ) : a flag referring to that part of the annex of double and multiple systems ( dmsa ) where the star has been classified , the component designations , the angular separation ( in arcsec ) , followed by the information about the variability : the periodicity , a flag indicating in which variability annex the star has been classified ( p/1=periodic variable ; u/2=unsolved ; m = possible microvariable ; d=duplicity - induced variability ; c=constant ) .
the final columns list the median hp magnitude with standard error and the differential hp magnitude with standard error
. + table [ tab : variability ] lists the name and the hd number ( or bd number if hd does not exist ) , remarks concerning the variability , the number of frequencies , the dominant periodicity and semi - amplitude , the projected rotational velocity of the @xmath0 sct star , the spectral types of the @xmath0 sct and additional components . among the remarks
the following notations are used : n = narrow range of frequencies ; w = wide range of frequencies ; r = radial pulsator ; nr = non - radial pulsator . [ tab : variability ] vw ari & 15165 & w & @xmath1 & p1=0.161 & a1=0.020 & 90 & f0iv & b = f2iv ' '' '' + bn hyi & 20313 & & 2 & 0.065 & 0.006 & 47 & f2ii - iii & - + v534 tau & 23567 & & & 0.032 & 0.009 & 90 & a9v & - + v647 tau & 23607 & & & 0.047 & 0.007 & 10 & a7v & - + v650 tau & 23643 & r+nr;w & 5 & 0.031 & 0.004v ? & 185 & - & - + v483 tau & 27397 & q=0.038 & & 0.0548 & 0.013 & 100 & f3v ... & - + v775 tau & 27628 & & & 0.06 & @xmath2 0.005 & 30 & a3 m & - + @xmath3 tau & 28024 & & & 0.148 & 0.010 & 195 & a8vn & - + v777 tau & 28052 & & & 0.182 & 0.010 & 195 & f0v ... & - + @xmath4 tau & 28319&n & @xmath5&p1=0.0756 & a1=0.007 & 140 & a7iii & b = g7iii + @xmath6 tau & 28910 & & & 0.07 & 0.005 & 103 & a8v & - + kw aur & 33959&q=0.012;n & 3&p1=0.0881 & a1=0.016 & 25 & a9iv & c = f3v + br cnc & 73175 & & @xmath73 & 0.040 & 0.010 & 170 & f0vn & - + bq cnc & 73729 & & 3 & 0.064 & 0.003 & 165 & f2v & - + @xmath3 uma & 84999&nr high _ m_&many & 0.07 - 0.09&@xmath80.03 & 115 & f0iv & - + ai com & 108662 & & 1 ? & 0.052 & 0.02 & 10 & a0p & a0p + gg vir & 110377 & & many & 0.023@[email protected] & @xmath20.014 & 160 & a7vn & g8v + @xmath9 boo & 124675&q=0.016;n & 4 & 0.0648 & 0.0064 & 140 & a8iv & f1v + @xmath10 boo & 125161&susp.@xmath0 sct & 1 ? & 0.026 & 0.0035 & 130 & a9v & - + v1208 aql & 181333 & q=0.034;n & @xmath11 & p1=0.146v&a1=0.013v & 48 & f0iii & g5 + - phe & 220392 & n & 2 & p1=0.2139 & a1=0.014 & 145 & f0iv & b = a9v + @xmath0 ser & 138917 & & 2 & 0.156&0.01 - 0.02 & 70 & f0iv & - + @xmath12 crb & 140436 & & @xmath11 & 0.111or0.296 & k=1 - 2 kms@xmath13 & 100 & a1vs & - + @xmath14 cyg & 202444 & & & 0.083@[email protected]&0.01 & 91 & f1iv & - + xz men & 31908 & & many & p1=0.1083 & a1=0.009 & - & fm ... & - + dg leo & 85040 & & @xmath15 ?
& 0.0826 & 0.015 & 18 & a8iv & am ? + dp uma & 104513 & & many & 0.04var&@xmath20.01 & 78 & a7 m & b = g8v + @xmath12 boo & 127762&nr high _ m_&2&0.073&0.050 & 135 & a7iiivar & - + wx eri & 21102&susp . @xmath0 sct & & 0.1645 & 0.010 & - & a7 & - + @xmath16 cep & 211336&nr high _ m_&@xmath11 ? &
p1=0.041 & a1=0.008 & 105 & f0iv & - + ai hya&+00 2259 & & & 0.1380 & 0.020 & 24 & f2 & - + @xmath17 tuc & 3112&unstable?;w & 10&p=0.049 & 0.013 & 80 & a7iv & - + cc and & + 41 119 & & 7 & 0.1249 & 0.24 & 20 & f3iv - v & - + ab cas & + 70 193 & q=0.036 & 1 & 0.0583 & 0.03@[email protected] & 55 & a3v & k1v + rz cas & 17138 & susp .
sct&many & 0.016 & 0.02 & - & a3v & k0iv + ry lep & 38882 & & & 0.2254 & 0.35 & - & a9v & - + v1004 ori & 40372 & & 1 ? & p1=0.054 & a1=0.01 & 71 & a5me & b = f7v + v474 mon & 40535&q=0.020 & many & p1=0.1361 & a1=0.15 & 19 & f2iv & - + uz lyn & 43378 & & & 0.065 ? & & 60 & a2vs & - + y cam & + 76 286 & & 1 & 0.0665 & 0.04v & 46 & a9iv & k1iv + ux mon & 65607 & & & 0.04 : & 0.04 & 37 & g2iiiv & - + sz lyn & 67390 & & 1 & 0.12 & 0.51 & 40 & a9 & f2 - 3 + rs cha & 75747 & & & 0.077 & @xmath20.01 & 65 & a8iv & a8iv + dl uma & 82620 & & & 0.0831 & 0.06 & - & f0 & - + kz hya & 94033 & & & 0.0595 & 0.8 & @xmath18 & - & - + fm vir & 110951&n & 1 & 0.0756v & .01-.04 & 24 & a8
m & a7 + v644 her & 152830 & & & 0.1151 & 0.040 & 20 & f5ii & - + @xmath19 ser & 160613 & & & & & 135 & a2va & - + v577 oph&+06 3679 & & & 0.0695 & 0.050 & - & a & - + @xmath0 sct & 172748 & r+nr & @xmath74 & p1=0.1938 & a1=0.29 & 30 & f2iiip & - + ce oct & 188520 & [email protected] & & p2=0.05451&a2=0.006 & - & a7iv / v & - + 18 vul & 191747 & & & 0.1215 & 0.008 & 45 & a3iii & - + @xmath0 del & 197461 & r ( comp b ) & 3@xmath20&0.1568 & 0.050 & 29 & a7iii & a7p @xmath0 del + ik peg & 204188 & & & 0.044 & 0.01 & 37 & a8 m & - + gx peg & 213534 & q=0.020;w & 5 & 0.056 & 0.015 & 46 & a5 m & - + @xmath21 cas & 432 & q=0.020 & 1 & 0.1010 & 0.015 & 69 & f2iii - iv & - + gn and & 2628 & q=0.017(r ) & 1 & 0.069 & 0.02v ? & 18 & a7iii & - + av cet & 8511 & & & 0.0685 & @xmath20.01 & 195 & f0v & - + v479 tau & 24550 & & & 0.0758 & 0.03 & - & f3ii - iii & - + v480 tau & 30780 & q=0.021 & & 0.04 & 0.02 & 145 & a7iv - v & g3v + ox aur & 50018 & & & 0.11544 & & 150 & f2v & g5v + @xmath6 pup & 67523 & r & 1 & 0.14088 & 0.09 & 15 & f2mf5iip & - + fz vel & 77140 & q=0.016 & & 0.065 : & 0.02 & 59 & am & - + rx sex & 90386 & & & 0.08 & @xmath20.01 & -&a3v & b = g5 + fg vir & 106384 & & & 0.0786 & 0.040 & 21 & am & - + @xmath22 lyr & 172167 & & & & & 5 & a0vvar & - + v1644 cyg & 192640 & & & 0.031 & 0.020 & 25 & a2v & b = g8iii ' '' '' +
in table [ table : vb ] we present additional information regarding the physical parameters of the components of some of the visual binaries listed in tables 1 and [ tab : variability ] given are the name and hd designation , the orbital period in years , the semi - axis major in au , the sum of the masses , and if known the radii @xmath23 , @xmath24 and the individual masses @xmath25 , @xmath26 as well as a reference . + we recall that the sum of the masses , @xmath27 ( in solar mass ) , and the orbital elements , p@xmath28 , the orbital period ( expressed in years ) , and a , the true semi - axis major ( in au ) , are linked through kepler s third law : @xmath29 there are three vb / o cases ( @xmath0 ser , @xmath12 crb , @xmath14 cyg ) but only two have a reliable orbit determination .
of course , only if the mass ratio is known from either the absolute astrometry or the spectroscopy can precise individual masses be obtained .
@xmath30 crb is not listed since it is very probably not a @xmath0 scuti star .
this orbital pair is in fact a triple system .
the visual binary has a period of @xmath8 1000 yr and a semi - axis major of 140 au .
it also contains a double - lined and chromospherically active spectroscopic binary . from differential spectrophotometry , frasca et al .
( 1997 ) deduce that there is * no * evidence for a 0.1 day periodicity but that all the variation is linked with the period of rotation .
they conclude that the photometric variability is due to dark spots on the secondary component of the sb and very probably * not * of @xmath0 scuti type .
this conclusion is also supported by the spectral classification ( f6+g0 ) .
we also discarded the following two visual double stars : v377 cas ( sep=2.1 ) because the light variations are not typical of the @xmath0 scuti type ( lowder 1989 ) and dl dra ( hr 5492 , sep=3.9 ) , shown to be probably constant already in 1990 ( papar et al . 1990 ) .
we next focus on some particular objects of this category . .@xmath0 scuti stars in visual binaries [ cols=">,>,^,^,^,^,^,^,^,^,^ " , ] here again , some misleading examples appeared in the literature .
for example ,
ry lep is among the list of southern eclipsing binaries observed by popper ( 1966 ) but he could only observe sharp single lines .
diethelm ( 1985 ) showed that this star is actually a relatively bright high - amplitude @xmath0 scuti star with a period of @xmath31 and an amplitude of 0.35 mag .
as reviewed by rodrguez et al .
( 1998 ) , ab cas is one of the clear examples where the light curves simultaneously and clearly show both types of variability : binarity and pulsation .
it is an algol - type binary system with a period of @xmath32 , while the primary component is a monoperiodic @xmath0 sct star with a pulsation period of @xmath33 .
the star is pulsating in the fundamental radial mode .
rodrguez et al .
( 1998 ) using a previously determined value for the mass ratio of 0.22 and assuming the secondary to fill the roche lobe , found the parameters quoted in table [ table : ecli ] .
together with ab cas this is one of the oldest @xmath0 sct variables known to be in an eclipsing system : its variability was discovered in 1903 by mrs .
its components ( a9 iv and k1 iv ) have rather different masses ( 1.9 m@xmath34 and 0.4 m@xmath34 ) but rather similar radii ( 3.15 r@xmath34 and 3.05 r@xmath34 , respectively ) .
the light curve is very reminiscent of the algol - type and the system is thought to be semi - detached , with the k star filling its roche lobe .
shapley ( 1917 ) , using the data of miss harwood , computed the first orbital elements of y cam , although a light curve was already obtained by nijland as quoted in the catalogue of shapley ( 1913 ) .
dugan ( 1924 ) obtained a complete visual light curve and computed a solution for the system .
it was shown that the dimensions of the two components were nearly the same .
szcepanowska ( 1955 ) , on the basis of 176 times of minimum , found it necessary to introduce a sinusoidal term into the elements of y cam which plavec et al .
( 1961 ) thought to be only the main part of the whole period variation .
they moreover concluded that the large periodic term could not be due to a third body nor to apsidal motion .
y cam nevertheless appeared in the catalogue of systems with apsidal motions of petrova & orlov ( 1999 ) , with an apsidal period of 60 years .
broglia & marin ( 1974 ) revealed the primary to be a @xmath0 sct star with a period of @xmath35 and a variable light amplitude .
the amplitude variation is not correlated with the phase in the orbital motion which led them to conclude that the `` amplitude variations in @xmath0 sct stars are not necessarily caused by a companion '' .
broglia & marin ( 1974 ) confirmed the variation of the orbital period , which they considered as a proof for mass transfer in the system .
does the binarity influence the pulsation properties of @xmath0 scuti stars ? in the light of the previous sections , it is obvious that there is no clear and easy answer .
every discussed case seems to be particular in its own way .
apart from the binarity , many other phenomena are involved that may even have a stronger impact on the pulsation properties of the stars : evolution , chemical composition , rotational effects ... in an effort to generalize , some authors consider the ratio between the pulsation period and the orbital period in order to search whether a resonance mechanism between both can occur ( e.g. frolov et al .
1980 , tsvetkov & petrova 1993 ) .
this is however only possible if both periods are determined with very high accuracy .
for example , in the case of sz lyn , the ratio between both periods is almost exactly 9800 ( taking the orbital period from moffett et al .
( 1988 ) ) . but with a slightly different value of the orbital period ( 1181.5 instead of 1181.1 ) ,
this ratio , 9803.26 , is not indicative .
such an exercise is therefore only applicable to very close systems for which tidal interactions are expected to be important ! we computed the ratio between the orbital and pulsation periods for the closest binary systems of our list ( with orbital periods up to 20 days ) and , in general , we do not find values close to an integer .
five systems however form an exception : kw aur , cc and , dl uma , v644 her and wx eri .
dl uma and wx eri have very short orbital periods ( 0.42 and 0.82 days , respectively ) and in both cases , this is almost exactly 5 times the pulsation period . for kw aur , cc and and v644 her ,
this ratio is 43 , 84 and 103 , respectively .
it is not clear whether these last values have any physical meaning at all .
+ in very close systems tidal forces may force the system towards synchronization . for spectral types typical of @xmath0 scuti stars
, this happens for binary systems with orbital periods of a few days up to 10 days if the stars are evolved .
synchronization will slow down the star .
thus , an indirect effect of binarity may be the change of the rotational velocity .
possible correlations between the amplitude of the pulsation and the rotation have been discussed before ( e.g. breger 1980 ; solano & fernley 1997 ) . such a correlation is also seen in the data presented here . + some cases present better evidence than others : e.g. amplitude modulation in the case of the radial pulsator @xmath17 tuc ( de mey et al . 1998 ) or possible frequency splitting in the case of the non - radial pulsator kw aur ( fitch & winiewski 1979 ) but other effects may also occur ( e.g. frequency shifts ) that are almost impossible to detect .
we need better models to check whether any of the above presented assumptions on the link between binarity and pulsation are valid and whether these can explain the observational facts in the most interesting cases .
nevertheless , we should carefully investigate @xmath0 scuti stars in binary systems from the observational point - of - view since they provide additional constraints on the physical parameters of the pulsating star , and therefore also on the characteristics of the pulsation .
for single field @xmath0 scuti stars , the position in the hr diagram and thus the evolutionary phase may be ambiguous , especially at the end of the core h burning phase , while @xmath0 scuti stars in binary systems can be located with much better accuracy .
it may also be especially worthwhile to study the _ differences _ in variability between two nearly identical components of a binary system , of which one or both may be @xmath0 scuti stars .
various such cases have appeared in the present discussions and many deserve further observations and study .
this research made use of the simbad database , operated at cds , strasbourg , france , as well as of the nasa astrophysics data system .
we also acknowledge funding by project g.00265.97 of the _ fonds voor wetenschappelijk onderzoek ( fwo , belgium)_. abt , h.a .
1965 , , 11 , 429 abt , h.a . ,
biggs , e.s .
1972 , in bibliography of stellar radial velocities , kitt peak national observatory abt , h.a . , levy , s.g . 1985 , ,
59 , 229 abt , h.a . ,
morrell , n.i .
1995 , , 99 , 135 adams , w.s .
1914 , , 26 , 261 adelman , s.j .
1997 , , 125 , 497 albitzky , v.a . 1933 , pulkova circ .
8 andersen , j. 1991 , astron.astrophys.review , 3 , 91 andrievsky , s.m .
, chernyshova , i.v . ,
usenko , i.a . ,
kovtyukh , v.v . ,
panchuk , v.e . , & galazutdinov , g.a .
1995 , , 107 , 219 ( ac95 ) baade , d. , bardelli , s. , beaulieu , j.ph . , vogel , s. 1993 , , 269 , 195 baize , p. & petit , m. 1989 , , 77 , 497 ( bp89 ) bardin , c. , imbert , m. 1981 , , 98 , 198 bardin , c. , imbert , m. 1982 , , 47 , 319 bardin , c. , imbert , m. 1984 , , 57 , 249 barnes , t.g . , moffett , t.j .
1975 , , 80 , 48 bartolini , c. , grilli , f. , parmeggiani , g. 1972 , ibvs 704 bartolini , c. , grilli , f. , parmeggiani , g. , piccioni , a. , silveri , p. 1983
, , 53 , 139 batten , a.h . , fletcher , j.m . , maccarthy , d.g .
1989 , publ .
victoria , 17 , 1 beardsley , w.r .
, zizka , e.r .
1977 , rev.mexicana de astro .
y astrofisica , 3 , 109 bertiau , s.j .
1957 , , 125 , 696 bolton , c.t .
, geffken n. 1976 , , 88 , 195 brancewicz , h.k .. , dworak , t.z .
1980 , acta astron . , 30 , 501 breger , m. 1970 , , 162 , 597 breger , m. 1973 , , 22 , 247 breger , m. 1979 , , 91 , 5 breger , m. 1980 , , 235 , 153 breger , m. , garrido , r. , huang , l. , jiang , s-.y . , guo , z .- h . ,
frueh , m. , paparo , m. 1989 , , 214 , 209 ( bg89 ) broglia , p. 1963 , mem.soc .
, 34 , 431 broglia , p. , marin , f. 1974 , , 34 , 89 budaj , j. 1996 , , 313 , 513 budaj , j. 1997 , , 326 , 655 caliskan , h. , adelman , s.j .
1997 , , 288 , 501 cannon , j.b . 1915 , pub.dom.obs.ottawa . 2 , 369 chaubey , u.s .
1983 , ibvs , 2348 chernyshova , i.v . ,
andrievsky , s.m . ,
kovtyukh , v.v . & mkrtichian , d.e .
1998 , contrib .
skalnat pleso 27 , 332 de mey , k. , daems , k. , sterken , c. 1998 , , 336 , 527 diethelm , r. 1985 , , 149 , 465 dommanget , j. , & nys , o. 1995 , bull .
dinf . du cds de strasbourg 46 , 3 dugan , r.s .
1924 , princeton contr .
no.6 duquennoy , a. & mayor , m. 1991 , , 248 .
485 eggen , o.j .
1956 , , 68 , 541 elliott , j.e .
1974 , , 79 , 1082 esa 1997 , the hipparcos and tycho catalogues , esa sp1200 fehrenbach , c. , duflot , m. , burnage , r. , mannone , c. , peton , a. , & genty , v. 1987 , , 71 , 275 fitch , w.s .
1967 , , 148 , 481 fitch , w.s .
1969 , , 158 , 269 fitch , w.s .
1976 , iau coll .
29 , p. 168
fitch , w.s . , & winiewski , w.z .
1979 , , 231 , 808 ( fw79 ) frandsen s. , jones a. , kjeldsen h. , viskum , m. , hjorth , j. , andersen , n.h . , & thomsen , b. 1995 , , 301 , 123 ( fj95 ) frasca , a. , catalano , s. , & mantovani , d. 1997 , , 320 , 101 frolov , m.s . ,
pastukhova , e.n . ,
mironov , a.v . , moshkalev , v.g .
1980 , ibvs 1894 frolov , m.s . ,
pastukhova , e.n . ,
mironov , a.v .
1982 , ibvs 2117 frost , e.b .
1924 , , 60 , 319 fu , j .-
n . , jiang , s.y .
1995 , , 110 , 303 garca , j.r . ,
cebral , j.r . ,
scoccimarro , e.r .
, wahnon , p. , arena , r. , bazterra , v. , pellizza , l. , risi , a. , rodr]iguez , m.l . ,
zimmermann , m. 1995 , 109 , 201 goupil , m.j . , michel , e. , lebreton , y. , baglin , a. 1993 , , 268 , 546 grenier , s. , burnage , r. , faraggiana , r. , gerbaldi , m. , delmas , f. , gomez , a.e . ,
sabas , v. , & sharif , l. 1999 , , 135 , 503 harper , w.e .
1933 , pub .
obs . , 6 , 203 harper , w.e .
1938 , pub .
6 , 207 hartkopf , w.i . ,
mcalister , , h.a . , & mason , b.d . 1999 ,
chara contrib .
4 , georgia state university , atlanta hartkopf , w.i . , & mcalister , , h.a .
1989 , , 98 , 1014 hillenbrand , l.a . , massey , ph . , strom , s.e . , & merrill , k.m .
1993 , , 106 , 1906 hoffmeister , c. 1949 , astron .
12 , 1 hummel , c.a . &
armstrong , j.t .
1992 complementary approaches to double and multiple star research , asp conference series , vol .
mcalister & w.i .
hartkopf , p.552 kaye , a.b . 1999 ,
ibvs , 469 kennelly , e.j .
& walker , g.a.h .
1996 , 108 , 327 ( kw96 ) kennelly , e.j . , walker , g.a.h . ,
catala , c. , foing , b.h . ,
huang , l. , jiang , s. , hao , j. , zhai , d. , zhao , f. , neff , j.e . , houdebine , e.r . ,
ghosh , k.k . , charbonneau , p. 1996
, , 313 , 571 king , j. , liu , t. 1990 , , 102 , 328 krlikowska , m. 1992 , 260 , 183 kurtz , d.w .
1980 , , 193 , 51 kurtz , d.w .
, breger , m. , evans , s.w . , sandmann , w.h .
1976 , , 207 , 181 lampens , p. , & van camp , m. 1999 , this volume ( lv99 ) lehmann h. , scholz , g. , hildebrandt , g. , & yang , s. 1997 , , 327 , 167 levato , h. 1976 , , 203 , 680 li , z .- p .
, zhou , a .-
, & yang , d. 1997 , 109 , 217 lindegren , l. 1997 , esa sp-402 , 13 liu y.y . , jiang , s.y . , michel , e. , hernndez , m.m . , chevreton , m. , auvergne , m. , mangeney , a. , belmonte , j.a . , goupil , m.j . , baglin , a. , roca corts , t. , & vidal , i. 1996 , , 120 , 179 lopez de coca , p. , costa , v. , rolland , a. , & walker , e.n . 1979 , , 36 , 61 lowder , w.m . 1989 ,
javso 18 , 116 lucy , l.b . , sweeney , m.a .
1971 , , 76 , 544 macnamara , d.h .
1997 , , 109 , 1221 mellor l.l . 1917 , pub .
mich . 3 , 61 mitton , j. , stickland , d.j .
1979 , , 186 , 189 michel , e. , et al .
1992 , , 255 , 139 michel , e. , hernndez , m.m . ,
houdek , g. , goupil , m.j . ,
lebreton , y. , prez hernndez , f. , baglin , a. , belmonte , j.a . , & soufi , f. 1999 , , 342 , 153 moffett , j.j .
, barnes iii , t.g . , fekel , f.c.jr , jefferys , w.h . ,
achtermann , j.m .
1988 , , 95 , 1534 morris , s.l . 1985 , ,
295 , 143 pan , x. , shao , m. , colavita , m.m . 1992 complementary approaches to double and multiple star research , asp conference series , vol .
mcalister & w.i .
hartkopf , p.502 papar m. , jiang s .- y . , li z .- p . , kollath z. 1990 , ibvs 3541 petrie , r.m .
1950 , pub .
victoria , 8 , 319 petrova , a.v .
, , orlov , v.v .
1999 , , 117 , 587 plavec , m. , smetanova , m. , pekny , z. 1961 , bull .
czechoslovakia , 12 , 117 popper , d.m .
1966 , , 71 , 153 reimers , d. 1976 , , 53 , 377 riboni e. , poretti e. , & galli g. 1994 , , 108 , 55 rodrguez , e. 1999 , , 111 , 709 rodrguez , e. , rolland , a. , lopez de coca , p. , garrido , r. , & gonzalez - bedolla , s.f .
1992 , , 96 , 429 rodrguez , e. , rolland , a. , lopez de coca , p. , martin , s. 1996 , , 307 , 539 rodrguez , e. claret , a. , sedano , j.l . ,
garca , j.m . , garrido , r. 1998 , , 340 , 196 roman , n.g . ,
morgan , w.w . , eggen , o.j .
1948 , , 107 , 107 romanov , yu . s. , fedotov , yu . t. 1979 , astron.tsirk .
1071 rosvick , j.m . , & scarfe , c.d .
1991 , , 103 , 628 scholtz , g. , lehmann , h. , hildebrandt , g. , panov , k. , iliev , l. 1998 , , 337 , 447 seeds , m.a . , yanchak , g.a . 1972 ,
ibvs , 625 shapley , h. 1913 , , 38 , 158 shapley , h. 1917 , , 46 , 56 slettebak , a. 1954 , , 119 , 146 smith , m.a . 1982 , , 254 , 242 sderhjelm , s. 1999 , , 341 , 121 solano , e. , fernley , s. 1997 , , 122 , 131 soliman , m.a . ,
hamdy , m.a . ,
szeidl , b. , szabados , l. 1986 , commun .
konkoly obs . hung .
sci . , budapest , no.88 struve , o. , sahade , j. , zebergs , v. 1957 , , 125 , 692 szatmry , k. 1990 , j. aavso , 19 , 52 szcepanowska , a. 1955 , acta astron . , 6 , 134 tokovinin , a. 1997 , , 121 , 71 torres , g. , stefanik , r.p . , & and latham , d.w .
1997 , , 485 , 167 ( tsl97 ) tsvetkov , ts.g .
, petrova , ts.c .
1993 , , 203 , 257 van leeuwen , f. 1997 , esa sp-402 , 19 vivekananda - rao , p. , sarma , m.b.k . ,
agrawal , p.c . , & apparao , m.v.k .
1985 , ibvs , 2721 weiss , w.w . ,
1998 , , 338 , 919 yang , s. , walker , g.a.h . ,
fahlman , g.g . , & campbell , b. 1982 , , 94 , 317
|
we present a list of @xmath0 scuti stars in double and multiple systems , ranging from the very wide binaries to the very close ones such as spectroscopic and eclipsing systems including the optical visual pairs which are of no further use here .
our aim is to group the information from the binarity on the one hand and the pulsational characteristics on the other hand for as complete a sample as possible of @xmath0 scuti stars in stellar systems .
a selection of 18 well - documented cases , taking care that every type of binary is being represented , is discussed more extensively .
|
let us consider a gas of charged particles @xmath0 on the complex plane @xmath1 , confined individually by the external field @xmath2 and experiencing a coulomb pair repulsive interaction .
this corresponds to the probability distribution on @xmath3 with density proportional to @xmath4 where @xmath5 is a fixed parameter and where @xmath6 is a fixed smooth function .
we are mostly interested in asymptotics in @xmath7 of this particles system . to ensure the integrability for large enough @xmath8 , and following @xcite , it is convenient to assume that there exists some real number @xmath9 with @xmath10 such that @xmath11 . indeed ,
using the inequality @xmath12 valid for any @xmath13 , we get @xmath14 the factor @xmath8 in front of @xmath2 in the density is motivated by the remarkable formula @xmath15 where @xmath16 is the empirical distribution of the particles . indeed , following @xcite ( see @xcite for more general models ) , on the space @xmath17 of probability measures on @xmath1 equipped with the topology of narrow convergence ( i.e. the dual topology related to bounded continuous test functions ) , the functional @xmath18 is strictly convex , bounded from below with compact level sets , admits a unique minimizer @xmath19 , and the empirical distribution @xmath20 satisfies a large deviations principle for the weak topology at speed @xmath21 with good rate function @xmath22 . in particular , from the first borel - cantelli lemma , with probability one , @xmath23 weakly as @xmath7 . following ( * ? ? ?
* th.1.3 ) , it can be shown that @xmath19 has compact support when @xmath24 .
the coulomb gas is the spectrum of an ensemble of random normal matrices obtained by conjugating with an independent haar unitary matrix , see for instance @xcite . on the other hand , such a coulomb gas also describes the spectrum of some few non normal random matrix ensembles .
for instance , if @xmath25 is a random @xmath26 matrix with i.i.d .
complex gaussian entries @xmath27 of mean @xmath28 and variance @xmath29 then its eigenvalues are a coulomb gas with @xmath30 and @xmath31 , and in this case @xmath19 is the uniform distribution on the unit disc of @xmath1 , see @xcite . furthermore , if @xmath32 is an independent copy of @xmath25 , then the spectrum of @xmath33 is a coulomb gas with @xmath34 and @xmath35 , and in this case , @xmath19 is not compactly supported and is actually heavy tailed , see @xcite . in both examples ,
@xmath34 and @xmath2 is radially symmetric .
we suppose from now on that @xmath2 is radially symmetric in the sense that @xmath36 where @xmath37 is smooth and such that either @xmath38 is increasing or @xmath39 is convex , with @xmath40 as @xmath41 .
let @xmath42 be the smallest number such that @xmath43 for all @xmath44 , and let @xmath45 be the smallest solution of @xmath46 .
then @xmath47 and , following ( * ? ? ?
6.1 ) , the probability measure @xmath19 is , in polar coordinates ( @xmath48 ) , @xmath49}(r){\mathbf{1}}_{[0,2\pi]}({\theta})\,drd{\theta}.\ ] ] this formula comes in fact from @xmath50 and @xmath51 .
let @xmath0 be the coulomb gas ( in other words the atoms of @xmath52 ) and @xmath53 be their reordered moduli ( in other words the order statistics of the moduli ) .
in particular , @xmath54 following rider @xcite ( see also @xcite ) , in the case @xmath34 and @xmath55 then @xmath56 as @xmath7 and the asymptotic fluctuation follows a gumbel law .
the aim of this note is to show , still for @xmath34 , the universality of this result beyond the quadratic case on @xmath39 .
the following structural result forms the basement of all the remaining results of this note .
[ th : layers ] if @xmath30 then we have the identity in distribution @xmath57 where @xmath58 are as in and where @xmath59 is the order statistics of independent random variables @xmath60 with @xmath61 of density proportional to @xmath62 for every @xmath63
. in other words , in terms of point processes this means @xmath64 theorem [ th : layers ] follows from the approach of kostlan @xcite used in the gaussian case @xmath55 ( see also ( * ? ? ?
* theorem 4.7.1 ) for a determinantal point processes point of view ) .
let us quickly recall the proof for convenience , which is elementary .
one starts from the vandermonde determinant @xmath65 hence , the density [ eq : cougas ] rewritten in polar coordinates ( @xmath66 ) is proportional to @xmath67 at this step , we observe that if @xmath68 with @xmath69 for some @xmath70 then @xmath71 consequently , we obtain @xmath72^n}\ ! \prod_{1\leq
j < k\leq n}|z_j - z_k|^2\,d{\theta}_1\cdots d{\theta}_n = ( 2\pi)^n\sum_{{\sigma}\in s_n}\prod_{j=1}^nr_j^{2({\sigma}(j)-1 ) } = ( 2\pi)^n\mathrm{per}{{{\left[r_j^{2(k-1)}\right]}}}_{1\leq j , k\leq n}\ ] ] ( `` @xmath73 '' stands for `` permanent '' ) and thus , the density of the moduli is proportional to @xmath74}}}_{1\leq j , k\leq n}\prod_{j=1}^nr_j = \mathrm{per}{{{\left[r_j^{2k-1}e^{-nv(r_j)}\right]}}}_{1\leq j , k\leq n } = \mathrm{per}{{{\left[f_k(r_j)\right]}}}_{1\leq j , k\leq n}.\ ] ] up to a factor @xmath75 , this is the density of @xmath76 where @xmath60 are independent random variables with @xmath61 of density proportional to @xmath77 for every @xmath63 , and where @xmath78 is a random uniform element of @xmath79 , independent of @xmath60 .
beware that the phases are not independent in theorem [ th : layers ] .
however , since the gas is rotationally invariant , we may deduce that the equilibrium measure @xmath19 exists , is rotationally invariant , and we may compute its radial part using the law of the moduli . if @xmath55 then @xmath80 is @xmath81-distributed with @xmath82 degree of freedom .
theorem [ th : layers ] allows to reuse the strategy behind the work of rider @xcite in the case where @xmath39 is a power .
the following theorem is proved in section [ se : power ] .
the case @xmath83 is the one of rider @xcite .
[ th : power ] let @xmath84 be as in , with @xmath34 .
suppose that @xmath85 for any @xmath86 , for some @xmath87 .
set @xmath88 and @xmath89 then @xmath90 converges in distribution as @xmath7 to the standard gumbel law : @xmath91 in particular @xmath92 converges in probability to @xmath93 as @xmath7 , in other words @xmath94 our next result below , which is proved in section [ se : genecase ] , goes beyond the power case considered in theorem [ th : power ] ( which corresponds formally to the special choice @xmath85 ) .
[ th : genecase ] let @xmath84 be as in , with @xmath34 .
additionally , let us assume the following properties on the potential @xmath39 : * * ( a1 ) * @xmath39 is strictly convex : there exists @xmath95 such that @xmath96 * * ( a2 ) * for each @xmath97 $ ] , there exists a unique @xmath98 such that @xmath99 let @xmath100 be the unique solution to the equation : @xmath101 .
let us define the sequences @xmath102 where @xmath103 and @xmath104 then @xmath105 converges in distribution as @xmath7 to the standard gumbel law : @xmath106 in particular , @xmath92 converges in probability to @xmath100 as @xmath7 , in other words , @xmath107 if the confining potential is not strong enough , the asymptotic fluctuation of @xmath108 is no longer gumbel .
for instance , in the case where @xmath109 for @xmath110 , we get from theorem [ th : layers ] that @xmath61 has density proportional to @xmath111 . in this case
, the muduli of the particles are heavy tailed , and the equilibrium measure @xmath19 exists and is heavy tailed .
one may wonder if the largest particle in modulus @xmath108 has frchet type fluctuations as @xmath7 .
recall that a random variable @xmath112 follows the frchet law of parameter @xmath113 iif @xmath114 for any @xmath115 .
in particular , @xmath116 as @xmath117 . at least formally , if one takes @xmath118 then @xmath119 and thus for every @xmath120 @xmath121 which gives @xmath122 and therefore @xmath123 the phenomenon is that a sum behaves asymptically like its largest term , which implies for the particles that the largest in modulus behaves like the one of largest index .
the edge universality remains untouched for general @xmath124-ensembles on the complex plane and for the eigenvalues of general random matrices with i.i.d.entries ( spectral radius ) .
the complex ginibre ensemble is exactly solvable and belongs to both categories .
the method used in the proof of theorem [ th : layers ] is maybe still usable when @xmath124 is an even integer .
one may also use it for gases in @xmath125 , @xmath126 , with density proportional to @xmath127 , in order to obtain the law of the norms of the particles ( by integrating non radial variables , possibly via a sort of wick formula on spheres ) .
a study of the bulk universality beyond logarithmic repulsion is considered in @xcite . on the other side ,
it is conjectured that the spectral radius of square random matrices with i.i.d .
centered entries of variance @xmath29 and finite fourth moment has also a gumbel type asymptotic fluctuations as @xmath7 , making the ginibre case universal again .
one may ask about the universality of refined aspects of the complex ginibre ensemble , such as the order statistics of the moduli of the eigenvalues , studied by rider in @xcite .
one may ask about the fluctuation of the smallest particle in modulus @xmath128 instead on the largest @xmath108 . for a general radially symmetric @xmath2 , the equilibrium measure
@xmath19 is radially symmetric but might be supported by more than one ring ( lack of radial connectivity ) .
one may then study the fluctuation at each edge ( inner and outer ) of these rings .
another natural question is to ask about the fluctuation at the edge in the single ring theorem @xcite .
this note benefited from the comments of two anonymous reviewers .
from theorem [ th : layers ] , @xmath129 has density proportional to @xmath130 which is @xmath131 , for every @xmath63 .
this gives the identity in distribution @xmath132 where @xmath133 are i.i.d .
of law @xmath134 ( mean and variance both equal to @xmath135 ) .
thus , for any deterministic sequences @xmath136 and @xmath137 in @xmath138 , and every @xmath139 , @xmath140 let @xmath141 lemma [ le : rider ] gives now that for @xmath142 and for every @xmath139 , @xmath143 it remains finally to use lemma [ le : delta ] with @xmath144 and @xmath145 ( one has also to use the slutsky lemma to obtain a bit nicer shift parameter @xmath146 ) .
we adapt the argument used by rider @xcite in the case of exponential random variables .
the driving intuitive idea is based on the law of large numbers and the central limit theorem , which indicate that the probabilities under the product are either asymptotically @xmath157 or identical to a gaussian deviation probability , leading to the maximum of independent gaussians , which is known to be gumbel .
namely , let @xmath158 and @xmath159 be deterministic sequences such that @xmath160 and @xmath161 , and let @xmath162 be the increasing function in both @xmath8 and @xmath163 given by @xmath164 the choices of @xmath165 and @xmath166 come from the following equivalence as @xmath7 @xmath167 from theorem [ th : layers ] we have @xmath168 we will show that the product over @xmath169 terms goes to @xmath170 as @xmath7 while the product over @xmath171 tends to @xmath157 as @xmath7 .
namely , denoting @xmath172 we get , @xmath173 let now @xmath174 be chosen large enough .
we can assume that @xmath8 is large enough so that @xmath175 using quantitative central limit theorem ( edgeworth expansion @xcite ) and assuming @xmath176 , one has that for a given @xmath177 $ ] @xmath178 the last term stems from hoeffding s concentration inequality to bound from above the probability that @xmath179 .
indeed , first one may use ( ii ) to get @xmath180 next , when @xmath70 runs from @xmath157 to @xmath181 and when @xmath182 and @xmath183 , one may see @xmath184 as an interpolation between @xmath185 and @xmath186 . by copying the arguments of @xcite , which essentially amounts to the riemann sum approximation of an integral , we obtain @xmath187 with our choice of @xmath185 , we get (
* eq . ( 15 ) ) that uniformly on compact sets in @xmath163 @xmath188 we now examine the contribution of the remaining terms : @xmath189 as the product is never larger than @xmath157 , one only needs to get a lower bound .
now , by ( i ) , @xmath190 for any @xmath70 such that @xmath191 it is not difficult to see that @xmath192 where the constant @xmath193 depends on @xmath194 all the difficulty lies in the regime where @xmath195 in this case we can use the moderate deviation result of ( * ? ?
* th . 5.23 p. 189 ) to get that for any @xmath196 @xmath197 actually we can refine the estimate
when @xmath198 @xmath199 combining the whole , we deduce that @xmath200 which obviously goes to @xmath157 as @xmath201 [ le : delta ] let @xmath202 be a sequence of random variables on @xmath138 and let @xmath203 be a @xmath204 function such as @xmath205 with @xmath206 . if there exist deterministic sequences @xmath136 and @xmath137 in @xmath138 such that we may use a taylor formula ( or intermediate value theorem ) to get @xmath215 where @xmath216 is a random variable lying between @xmath166 and @xmath217 .
thanks to ( i ) , ( ii ) , and the slutsky lemma , we have @xmath218 in law , and thus in probability because the limit is deterministic .
therefore @xmath219 in probability , and thus , by ( iii ) , @xmath220 in probability .
since @xmath221 is continuous at point @xmath222 , the continuous mapping theorem gives that @xmath223 in probability , and thus , by ( iv ) , @xmath224 in probability .
now it remains to use ( i ) and the slutsky lemma to obtain @xmath225 assumption * ( a1 ) * can in principle be relaxed to the strict convexity of @xmath237 for all @xmath238 the asymptotic expansion can then be performed unchanged provided that there exists @xmath95 so that for any such @xmath70 @xmath239 recall that @xmath240 is distributed as the order statistic @xmath241 of @xmath242 which are independent random variables with @xmath243 of density proportional to @xmath244 we shall first consider the value ( if it exists ) around which the random variables @xmath243 concentrate for a large @xmath245 typically if @xmath246 , we here show that @xmath243 concentrates around the deterministic number @xmath247 indeed , if @xmath248 be a bounded continuous function @xmath249 , then we consider the asymptotic expansion of @xmath250 when @xmath251 using the laplace method .
we have @xmath252 where @xmath253 to determine critical points , one uses that @xmath254 by assumption * ( a2 ) * , there exists a unique solution @xmath255 to the equation @xmath256 and by * ( a1 ) * one has that @xmath257 for some constant @xmath258 ( independent of @xmath259 ) . also by * ( a1 )
* this critical point is non degenerate as @xmath260 note that this readily implies the gaussian decay of the exponential term : for all @xmath261 , @xmath262 by a standard laplace approximation , one then deduces that for any @xmath263 and for any @xmath264 such that @xmath265 the following holds : @xmath266 as a consequence , if @xmath264 is very large , that is those integers @xmath264 such that @xmath267 using that @xmath268 we conclude that the @xmath243 s concentrate around @xmath269 the constant @xmath100 will play the role of @xmath270 in the previous analysis .
let us now try to give the main ideas of the rest of the proof .
let @xmath271 be a given real number .
using theorem [ th : layers ] , in order to determine the limiting distribution of @xmath108 we consider the following : @xmath272 where @xmath273 is growing to infinity with @xmath8 and to be determined .
consider first @xmath274 for some @xmath275 that we choose as @xmath276 the value of the real @xmath277 will be fixed later .
one can first observe that @xmath278 for some @xmath279 .
for each @xmath280 one has that @xmath281 hereabove we have used the asymptotic expansion of @xmath282 using the laplace method , and the fact that @xmath283 for the error control , and in the last line we made the change of variables @xmath284 .
assume now that @xmath285 because @xmath286 , the integral in is of order @xmath287 .
we need to refine this rough order . to that aim
, we need to control the variation of @xmath288 ( and the gaussian decay ) in a bounded neighborhood of @xmath289 .
let @xmath290 be given ( small ) , that we fix hereafter .
let us first examine the decay of @xmath288 `` far '' from @xmath289 . using the upper bound @xmath291 forall @xmath292
, one can see that only a neighborhood of @xmath289 of width @xmath293 can contribute to the integral .
we then turn to a neighborhood of @xmath289 .
@xmath294 one can choose @xmath295 small enough so that for any @xmath296 @xmath297 thus using the change of variables @xmath298 and by a straightforward taylor expansion , we deduce that for all @xmath280 , @xmath299 the above estimate on the error @xmath300 is not precise enough as , later , one will sum this error over @xmath301 integers @xmath302 .
we need to refine the above estimate .
[ le : base ] if @xmath303 is @xmath304 function such that @xmath305 and @xmath306 then there exists @xmath307 such that for all @xmath308 , @xmath309 where @xmath310 for some constants @xmath311 and @xmath312 .
first , one can write @xmath313 from this , one deduces that @xmath314 set now @xmath315 then again @xmath316 using that @xmath317 one then deduces that @xmath318 thus we get that , every @xmath307 and for all @xmath292 such that @xmath319 @xmath320 as @xmath321 one then deduces that for @xmath295 small enough and any @xmath308 , @xmath322 this yields the desired estimate . back to the proof of lemma [ le : genecase ] , thanks to lemma [ le : base ] , can be improved to @xmath323 note first that the above estimates hold uniformy in @xmath280 provided that @xmath324 as @xmath8 goes to infinity .
in particular the constants @xmath325 do not depend on @xmath326 note also that the same estimate holds if one multiplies the integrand by the indicator function @xmath327 let now @xmath328 be such that @xmath329 as @xmath8 grows to infinity .
then , using the fact that the third derivative @xmath330 is bounded uniformly in @xmath264 when @xmath331 , we deduce that @xmath332 indeed the integral can be explicitly computed .
lastly one has that there exists a constant @xmath333 such that for any @xmath280 @xmath334 combining the whole yields that up to an error term ( of order @xmath335 ) one can replace the exponential term by the gaussian one obtained by the taylor expansion .
let @xmath174 be some large positive real number .
here we assume that @xmath336 and prove uniform convergence on compact sets of the c.d.f .
: one has that @xmath337}}}\right\ } } } } , \label{eq : majoerreur } \end{gathered}\ ] ] where @xmath338 then the sum over @xmath264 of the error term in can be bounded from above by @xmath339 .
this error term is negligible provided @xmath340 to sum up , we have @xmath341 where we have noted @xmath342
let us now choose @xmath343 then @xmath344 provided @xmath345 is bounded from below and one can easily check that with such a choice of @xmath248 all the required estimates on @xmath346 hold true . also there exists a constant @xmath347 such that for any @xmath280 @xmath348 the constant @xmath258 can be expressed in terms of @xmath100 and the derivatives of @xmath39
: it may be zero .
the important fact is that @xmath349 using now the same arguments as in @xcite for the riemann sum approximation , we again deduce that uniformly on compact subsets , @xmath350 now there remains to consider the smaller integers @xmath264 .
assume now that @xmath351 .
the exact critical point @xmath289 depends on @xmath8 and might now be closer and closer to @xmath352 where @xmath353 .
we need to control the second derivative of @xmath288 to be able to undertake the same laplace analysis as above .
@xmath354 by the convexity assumption , we deduce that @xmath355 so that the second derivative is negative and bounded from above in the whole interval @xmath356.$ ] from that one can deduce that @xmath357 by the strict convexity of @xmath39 .
in particular for all @xmath358 one has that @xmath359 with @xmath360 for some constant @xmath361 .
thus we deduce that @xmath362}}}\\ & \geq \prod_{i=0}^{i_0 } { { { \left[1- ( \sqrt{n}(1/2+o(1))\int_{t_0+\frac{g_n(u)}{\sqrt n}}^{\infty}e^{n f_i(t)-n f_i(t_i)}dt\right]}}}\\ & \geq \prod_{i=0}^{i_0 } { { { \left[1-(\sqrt{n}(1/2+o(1))e^{-n a ( t_0+\frac{g_n(u)}{\sqrt n}-t_i)^2}\right]}}}\\ & \geq \prod_{i=0}^{i_0 } { { { \left[1-(\sqrt{n}(1/2+o(1))e^{-ac_5c^2\log(n)}\right]}}}. \end{aligned}\ ] ] let us now choose @xmath363 ( determining by this way @xmath275 ) so that @xmath364 the latter product goes to @xmath157 as @xmath365 goes to infinity .
this is the needed estimate .
j. ben hough , manjunath krishnapur , yuval peres , and blint virg , _ zeros of gaussian analytic functions and determinantal point processes _ , university lecture series , vol .
51 , american mathematical society , providence , ri , 2009 .
mr 2552864 ( 2011f:60090 ) valentin v. petrov , _ limit theorems of probability theory _
, oxford studies in probability , vol . 4 , the clarendon press oxford university press , new york , 1995 , sequences of independent random variables , oxford science publications .
mr 1353441 ( 96h:60048 ) dnes petz and fumio hiai , _ logarithmic energy as an entropy functional _ , advances in differential equations and mathematical physics ( atlanta , ga , 1997 ) , contemp .
217 , amer .
soc . , providence , ri , 1998 , pp . 205221 .
mr 1606719 ( 99b:46101 ) edward b. saff and vilmos totik , _ logarithmic potentials with external fields _ , grundlehren der mathematischen wissenschaften [ fundamental principles of mathematical sciences ] , vol .
316 , springer - verlag , berlin , 1997 , appendix b by thomas bloom .
mr 1485778 ( 99h:31001 )
|
we consider in this note a class of two - dimensional determinantal coulomb gases confined by a radial external field . as
the number of particles tends to infinity , their empirical distribution tends to a probability measure supported in a centered ring of the complex plane .
a quadratic confinement corresponds to the complex ginibre ensemble . in this case
, it is also already known that the asymptotic fluctuation of the radial edge follows a gumbel law .
we establish in this note the universality of this edge behavior , beyond the quadratic case .
the approach , inspired by earlier works of kostlan and rider , boils down to identities in law and to an instance of the laplace method .
|
the existence of the knee in the spectra of extensive air showers ( eass ) has been known by now for more than 40 years @xcite .
first seen in the number of electrons ( the shower size ) observed near sea level it was later also observed in the muon number @xcite , hadron properties @xcite and muon densities @xcite .
in fact it seems to show up in all shower observables if investigated in sufficient detail .
nevertheless its origin is still obscure . of the explanations
proposed two seem to have found more general acceptance .
+ the first of these relates the knee to the influence of interstellar magnetic fields during propagation of the cosmic ray particles in or leakage from the galaxy or in the course of acceleration for which magnetic fields probably play a major role . since the radius of curvature of an extremely relativistic particle in a magnetic field is proportionate to its energy @xmath0 and inversely proportionate to its nuclear charge @xmath1 one would expect the knee then to show up at different energies for different nuclear species among the primary particles .
in fact one would expect the energy spectra of each element to show a knee at an energy displaced by a factor of @xmath1 with respect to that of protons .
this was probably first realized by peters @xcite who discussed the implications on various shower observables in great detail .
the effect of this on the total energy distribution of cosmic ray particles is illustrated in fig .
[ wiebsk ] which shows the result of a simple model calculation based on data compiled by wiebel - sooth et al .
these authors give , for each element up to nickel ( @xmath2 ) , the differential flux at 1 tev and the exponent of the power law spectrum .
these partial spectra were extrapolated up to a knee which was assumed to lie at @xmath3 ) = 3.4 + lg z$ ] .
this corresponds to a knee energy of c. 2.5 pev for protons .
it was then assumed that the exponent increased abruptly at the knee by 0.35 for all elements .
all statistical errors ( which are of course substantial owing to the extrapolation by some orders of magnitude ) were neglected . the full line in fig .
[ wiebsk ] represents the sum of all contributions from hydrogen to nickel .
the two dotted lines are the extrapolations of the all particle spectrum from below the proton and above the iron knee . as fig .
[ wiebsk ] shows the change of slope is , unsurprisingly , not confined to one position or small range along the line .
at least three bends are obvious corresponding to hydrogen , helium and the iron group .
hence in this case one would clearly expect a more complicated structure of the spectrum than a simple bend for all shower observables depending on primary energy .
+ the second proposal , less popular than the first one though recently reasserted by nikolsky and romachin @xcite , attributes the origin of the knee to the properties of high energy interactions in the atmosphere . a change of the spectrum of observables on ground level might of course occur if strong interaction changes by some kind of threshold phenomenon .
the knee energy is estimated to be near a few pev which is approximately a factor of 2 above the highest centre of mass energy available in the laboratory today .
therefore such an effect can not at present be excluded . since an eas induced by a nucleus of mass number @xmath4 and energy @xmath0 may , to a reasonable first approximation , be considered as a superposition of @xmath4 showers induced by nucleons of energy @xmath5 ( superposition principle ) one would expect , under this assumption , a similar shift of the knee position as described above but by a factor of @xmath4 instead of @xmath1 . again
a more complex structure of the knee would appear natural and a picture very similar to the one shown in fig .
[ wiebsk ] would be obtained .
+ it should be mentioned that the kascade collaboration has recently presented evidence that the knee is to be attributed to light nuclei and that the spectrum of heavier nuclei does not exhibit a change of slope in the vicinity of the main knee .
this claim is based on a comparison of electron and muon numbers of eass with monte carlo simulations @xcite but also on a phenomenological classification of eas by their electron to muon ratios @xcite .
+ erlykin and wolfendale have recently , in a series of papers @xcite , claimed observational evidence for a more complicated structure of the knee region .
this structure which according to the authors does not show up clearly in single measurements but becomes visible when averaging several ones , is attributed to the influence of a recent nearby supernova .
it is probably not unfair to say that not many have been convinced by the empirical evidence claimed by the authors .
but the underlying idea appears very intriguing and reasonable .
the solar system has , during the 4.5 billion years of its existence , probably been passed by several if not many shock fronts from supernovae exploding in its vicinity .
one of these clearly must be the most recent one and it appears well conceivable that the cosmic ray spectrum which we observe today is influenced by this individual event ( and hence not typical for the whole galaxy ) .
the possibility of a single source having a large impact on the energy and mass distributions of cosmic rays at the earth was already realized by peters @xcite .
+ for all these reasons it appears interesting to study the shape of the knee region in more detail and to look as to whether it can be really described by a simple bend between two power laws or whether any of the effects mentioned above can be identified . except for the work by erlykin and wolfendale
the author is not aware of any other attempt in this direction .
a definite negative result would not only worry some authors but also present difficulties for the usual models of the origin of the knee whose existence , on the other hand , is beyond any doubt .
+ in this paper i attempt to compare 28 different measured spectra of the electron number @xmath6 in the knee region .
( i drop the usual subscript @xmath7 because there can be no confusion in this paper . )
the data originate from 7 experiments and cover a range of atmospheric depths between 730 and 1250 [ @xmath8 .
the electron number ( or shower size ) is probably the shower observable for which the largest amount of measurements exist .
the basic procedure adopted is the following : each spectrum is first fitted separately by an adequate function adjusting either 4 or 5 parameters . in a next step the residual spectra
are shifted to the same knee position and averaged .
it may be expected that this averaging reduces not only the statistical fluctuations of the measurements but also ( at least part of ) the systematic ones and hence should make any deviations from the pre - chosen fit function more conspicuous . +
the data used and their analysis are described in the following section . in section 3
we compare the results of the analysis with a simple model and present our conclusions in section 4 .
a list of the spectra analysed in this paper and their sources can be found in table 1 .
the eas - top experiment is the only one which has published @xcite the @xmath6 spectra in numerical form instead of @xmath9 @xcite . ] .
the casa data were read from table 6.2 of reference @xcite ( which is equivalent to fig . of ref .
such a procedure is of course of limited accuracy and does not exhaust the statistical precision of the data ( especially at low shower sizes ) .
all other data sets were made available in numerical form by the authors .
table 1 gives some details of the data together with the quality of the two kinds of fits described in the next subsections .
several experiments registered events in different ranges of zenith distance which then of course correspond to different atmospheric depths .
the total data set comprised 784 points which is to be compared with a total of 112 or 140 derived parameters for the four or five parameter fits , respectively .
all fits performed were least squares fits with @xmath10 and @xmath11 as the variables and employing the usual weights derived from the quoted errors . here
@xmath12 is the differential flux . in a first step , the fit function chosen consisted of two power laws connecting continuously at the knee position : @xmath13 here @xmath14 is the differential flux at the knee position and @xmath15 for @xmath16 and @xmath17 for @xmath18 .
the quantities @xmath19 and @xmath20 were adjusted to describe the data . after the fit , the differences between observed and fit values were calculated for each of the 28 spectra .
these spectra of residuals were then shifted along the @xmath10 axis to the same knee position .
this amounts to choosing @xmath21 as the new independent variable where @xmath22 is the knee position found for the respective spectrum .
the scatter of the data points after these procedures is illustrated in fig .
[ diffig ] which clearly exhibits the statistical nature of the residuals ( and the increase of the errors with increasing @xmath6 ) .
no errors bars have been drawn because this would completely confuse the picture but most of them are compatible with 0 . + as the next step , all data points within horizontal intervals of width 0.1 were averaged neglecting the horizontal uncertainties resulting from the errors of @xmath22 .
the latter were smaller than the chosen bin width for almost all spectra .
the number of data points within a given interval was above 30 in the vicinity of the knee and dropped to near 1 at the extreme ends of the total range because the @xmath6 ranges of the various measurements did not coincide .
it should be mentioned that most of the original data were also binned in intervals of 0.1 width so this choice was very natural .
+ the averaging procedure described actually amounts to taking the geometric means of @xmath23 ( with adequate weights ) .
this eliminates all sensitivity to the absolute normalization of the data .
there is reason to suspect that the quantity which different experiments call shower size @xmath6 is not the same .
this is most obvious when the influence of muons on the experimental results is considered .
the electron detectors , in all experiments scintillation counters , are also sensitive to muons .
some experiments , such as kascade @xcite , put a lot of effort into correcting for this effect .
this is impossible , on the other hand , if muons are not measured independently as is the case in some of the experiments included in this analysis .
also different detector thresholds or saturation effects may lead to systematic differences .
the advantage of choosing @xmath21 as a new variable is that constant scale factors cancel .
although not all of the effects mentioned above will lead to modifications of @xmath6 by a constant factor at least some instrumental effects will be reduced or even removed .
+ the weighted averages thus obtained are displayed in fig .
[ dif4pmit ] with their statistical errors ( which have been multiplied , as usual , by the root of @xmath24 if the latter was larger than 1 ) .
a number of points at the upper end of the spectrum have been omitted here because of their huge errors . these were all compatible with 0 within their statistical errors . in spite of the considerable errors the mean residuals
can hardly be said to scatter statistically .
the most significant deviations , at the two points neighbouring the knee position , amount to @xmath25 and @xmath26 and are therefore highly significant . + a large discrepancy in the immediate vicinity of the knee position is in no way surprising because several experiments have shown that the slope of the spectrum does not change abruptly at the knee but rather smoothly over a finite range of @xmath10 .
the sign of the deviation is precisely as would be expected in such a situation .
so the most prominent deviation just reflects a well known and expected feature . on the other hand
, this result also shows that the procedure adopted in this paper is suitable to reveal more clearly a structure where it is present .
but obviously a more suitable fit function has to be employed .
+ it is worthwhile mentioning that the knee positions resulting from the fits show the well known dependence on atmospheric depth .
five of the seven experiments with 18 spectra are in good agreement with a simple linear dependence .
the other two can be made to agree when the shower sizes are scaled down by 30% and 70% .
the latter would not affect the present analysis in any way , of course .
the scatter of the other parameters is clearly larger than the quoted statistical errors would indicate .
this is subject of further analysis . since in this study
we are focussing on deviations from a smooth size dependence in a limited size range i would not expect such ( apparently systematic ) differences between experiments to influence my conclusions .
+ several different methods have been used to describe a smooth knee all of which seem to work reasonably well .
a very popular procedure is to select a finite interval in the vicinity of the knee and then first fit power laws to the regions above and below this interval . in a second step
these functions are then connected smoothly , e.g. by a polynomial .
a disadvantage of this method is , in my opinion , that at least two but usually more additional parameters are introduced ( which does not seem to be always realized ) .
+ ter - antonyan and haroyan @xcite have used a function which avoids this disadvantage .
their expression @xmath27 can be shown to approach pure power laws far from the knee and to tend to eq .
( 1 ) for @xmath28 .
the additional parameter @xmath29 does not lend itself to a direct physical interpretation , though .
also the two regions above and below the knee are not described in a formally equivalent way .
+ i have adopted a different approach .
it consists of folding eq .
( 1 ) with a normalized distribution of finite variance .
this implies a model in which a hypothetical underlying spectrum with a sharp knee is observed with finite resolution .
the exact shape of the resolution function is not really important but if a log - normal distribution is chosen the following closed expression is obtained : @xmath30 \left ( \frac{n_{k}}{n}\right ) ^{\gamma_{2}}\right)\ ] ] @xmath31 here @xmath32 is the standard deviation of the log - normal distribution and @xmath33 is the error integral : @xmath34 it is again straightforward to show that eq . ( 2 ) approaches power laws at sufficient distances from the knee and tends to eq .
( 1 ) for @xmath35 .
as far as the physical interpretation is concerned one should realize that @xmath32 does not primarily represent instrumental effects .
it is well known from simulations of eas development in the atmosphere that the shower size @xmath6 fluctuates considerably for events of fixed primary energy , mass and direction of incidence .
the effects of these fluctuations are also incorporated in the parameter @xmath32 .
+ i should like to point out that it is , in my opinion , very much a matter of taste which of the described approaches is chosen .
since all describe the experimental spectra well within observational uncertainties they will all lead to the same conclusions .
+ the fitting and averaging was then performed as described in the preceding subsection .
fig [ dif5pmit ] displays the result .
the residuals are now below 0.005 ( which corresponds to a difference between fit and measurement of less than @xmath36 ) for the whole lower part of the spectrum , up to @xmath37 .
+ the remaining deviations in the upper part amount to @xmath38 on the positive side and @xmath39 on the negative side . their regular pattern with increasing size can hardly be called statistical in spite of the fact that only two of the last 15 data points differ from 0 by more than 2 standard deviations .
a possible explanation is discussed in the next section where the origin of the dotted curve in fig .
[ dif5pmit ] will also be explained .
+ one might argue that several data points in the lower size range clearly differ from 0 by a similar amount statistically or even recognise an oscillatory structure .
but this deviation is on the 1 % level and hence a factor of 5 to 10 smaller than the one in the upper size range .
i would tend to argue , to the contrary , that an agreement between various experiments on this level is rather impressive and would hesitate to draw far reaching conclusions from these observations . +
the results displayed in fig . [ dif5pmit ] show clearly that the lower part of the observed size spectrum is well describeded by two power laws and a simple knee .
i find it difficult to believe , on the other hand , that the discrepancy visible in the upper part of the size range is purely statistical , in spite of the large errors of the individual points ( but this is of course left to the reader s own judgement ) .
it is much more difficult to assess possible systematic errors .
i will come back to the latter question later in this section and first turn to a possible explanation of the observed deviations assuming they are real .
+ let us suppose that the real size spectrum exhibits two knees displaced by one or two orders of magnitude with respect to each other .
this is not unreasonable in view of the considerations extended in the introduction and illustrated in fig .
[ wiebsk ] .
then a situation might occur which is sketched in fig .
[ schem ] . here
the full line represents a hypothetical spectrum with two knees .
if one tries to fit this spectrum with an expression allowing only one knee this will be adjusted to the lower one because it is more pronounced .
also statistical errors of the data points increase with increasing shower size and hence the lower points carry larger weights . but the deviation of the data from a pure power law in the higher size range will influence the fit and might lead to a situation illustrated by the dotted line in fig . [ schem ] .
this would then result in differences between data and fit which agree qualitatively with those observed in fig .
[ dif5pmit ] .
i admit that the separation of the full and dotted lines in fig .
[ schem ] does not look dramatic but one should be aware of the difference in the vertical scales between figs .
[ dif5pmit ] and [ schem ] .
+ it is possible to extend this conjecture a step further . for this
let us try to estimate the _ size _ spectrum resulting from an _ energy _ spectrum as displayed in fig .
[ wiebsk ] .
essentially all simulations of eass lead to a power law relation between shower size and energy for primary protons , at least in the range of energy and atmospheric depth relevant for this analysis : @xmath40 the corresponding relation for a heavier primary may then be estimated by help of the superposition principle invoked already in the introduction : @xmath41 here @xmath42 denotes the energy of a nucleus of mass number @xmath4 . assuming again , as in the introduction , that the knee position in energy is shifted by a factor of @xmath1 , @xmath43 , one obtains for the shift of the knee in shower size @xmath44 @xmath45
so in this simple model the shift depends only on mass and charge of the nuclei .
the second column of table [ knie ] shows these shifts for some abundant elements .
the value of @xmath46 which is the only additional information needed , was calculated from simulation results reported in tables 20 and 26 of @xcite ( for the qgs - jet model ; the other 4 models given in the same tables differ from this by at most 0.04 ) .
the separation between hydrogen and helium is reduced as compared to the corresponding shift in the energy spectrum due to the difference in mass to charge ratio .
so the knees of these elements would merge if the smoothness is taken into account .
hence in this case the knee observed in the size spectrum has probably to be attributed to the combined effect of protons and helium .
it is obvious that the shift of the knee for iron is of the right order of magnitude to explain a deviation as seen in fig .
[ dif5pmit ] .
+ .estimated shifts of the knee in the size spectrum with respect to the proton knee under the simple assumptions that the knee energy is either proportionate to charge or to mass of the primary nuclei . [ cols="^,^,^ " , ] if the other ( less popular ) proposal for the origin of the knee is adopted , @xmath47 , one obtains in a similar way @xmath48 the corresponding values have also been entered into table [ knie ] .
intriguingly , the position of the iron knee obtained under the assumption of @xmath49 is in better quantitative agreement with the deviations observed in fig .
[ dif5pmit ] . on the other hand , the knees originating from protons and helium
should then appear well separated of which there is not much evidence in the figure .
+ in order to check this possibility further i have calculated the following toy model .
i have assumed that the size spectrum can be described by the sum of two functions of the type given in eq .
( 2 ) with different intensities , the same slopes and the two knees separated by @xmath50 .
this model spectrum then has two free parameters to adjust , the ratio of intensities and the change of the power law exponents .
for the latter @xmath51 was assumed .
the parameters @xmath32 were chosen to reproduce the variance of the simulation results in the quoted tables of ref .
@xcite and were 0.44 for protons and 0.08 for iron .
this choice neglects all instrumental effects on the size resolution and should therefore represent a lower limit .
this calculated spectrum was then fitted by a single function of the same type with the knee at the same position as the lower of the other two .
intensity and slope below the knee were taken to be the same as those of the model spectrum , and the slope above the knee adjusted .
the difference of the model and fit spectra are shown in fig .
[ dif5pmit ] as the dotted line .
it should be emphasized that the width of the knees were not varied but kept fixed at the values quoted above .
the three new parameters ( relative intensity , change of the exponent of the model function and slope above the knee of the fit function ) were adjusted to some extent but no serious attempt was made to obtain a perfect fit .
for this all elements expected to be present in primary cosmic rays should have to be taken into account .
also there is no compelling reason to believe that all partial spectra have the same exponents .
this would then leave more parameters to adjust than data points in fig .
[ dif5pmit ] .
although the dotted line does not reproduce the data perfectly the model gives a reasonable description in view of its crudeness .
hence the data can be said to be in agreement with a second component in the overall spectrum exhibiting a knee at higher shower size .
it remains surprising , though , that the displacement is clearly in better agreement with a proportionality of the knee position to mass rather than to charge of the primary nuclei ( if the second component is attributed to iron which appears to be the most reasonable choice ) .
+ this is probably the right moment to stop speculating and to turn to the thornier question of systematic errors .
it should be mentioned that the number of experimental points averaged drops from 29 at @xmath52 to 7 at 2.1 .
so the region where the deviation is observed covers the highest data points of several experiments and these might be under suspicion to suffer from saturation effects .
saturation will lead to an overestimate of the differential flux increasing towards the end of the range of measurement .
therefore it is , in my opinion , not possible to exclude at this moment systematic effects as a possible origin of the observed deviations without more detailed investigations .
these require a better knowledge of experimental details and have to be performed for each experiment separately .
this is beyond the scope of the present paper .
i have presented an analysis of 28 size spectra from 7 different experiments .
their results can be well represented by two power laws with a smooth knee in between up to a shower size a factor of @xmath53 above the bend . at larger sizes a deviation occurs which appears statistically sound though not overwhelming .
the maximum of this deviation roughly coincides with the position where one would expect a knee originating from primary iron nuclei if the knee in energy is assumed to be shifted in proportion to primary mass .
i realize of course that , to the best of my knowledge , no such feature has ever been seen before in _ size _ spectra although a second knee was seen in the derived _ energy _
spectrum by the akeno group @xcite .
( the deviations claimed by erlykin and wolfendale @xcite are closer to the main knee position and can not therefore be identified with the ones observed here . )
but on the other hand no analysis so far has taken advantage of the statistics of 28 spectra from 7 experiments .
the effect shows up in a region where detectors might start to saturate which may lead to distortions of the spectrum .
hence systematic errors can not , in my opinion , presently be ruled out as the origin of the observed fine structure .
+ i would consider it worthwhile to study the data from experiments with high statistics and high detector quality in more detail , especially as far as saturation effects are concerned , in order to verify or not the existence of this fine structure .
one way of checking the influence of saturation is to compare spectra taken by the same experiment in different bins of zenith distance .
the measured size of a primary of given energy and mass decreases with increasing zenith angle .
this results in a shift of the knee position to lower sizes with increasing zenith distance .
this shift has been observed by many experiments and is also clearly visible in the data of the 5 experiments which have contributed more than one spectrum to this investigation .
hence vertical showers should saturate at lower values of @xmath54 .
if the same structure is observed at all zenith distances the influence of saturation can be ruled out .
the range of shower sizes of such a study should of course extend up to at least two orders of magnitude above the main knee .
this implies a reduction of particle flux by about 4 orders of magnitude and hence requires very good statistics as well as detectors of a considerable dynamic range .
+ it should be mentioned that the analysis described in section 2 of course also yields numerical values for the 4 or 5 parameters and their statistical errors and it appears interesting to compare the results from the various experiments and study their dependence on atmospheric depth .
this will be the subject of a forthcoming paper .
+ * acknowledgements * + + writing this paper would have been impossible without the generous support of quite a number of colleagues .
i thank especially a. chilingarian ( yerevan ) , r. glasstetter ( karlsruhe ) , g. heinzelmann ( hamburg ) , n. n. kalmykov ( moscow ) and m. nagano ( tokyo ) for making the results of their respective experiments available in numerical form .
i am also grateful to p. grieder ( bern ) and l. jones ( ann arbor ) for their help in retrieving literature and d. heck and m. roth ( both at karlsruhe ) for their advice concerning latex and paw ( and their patience ) .
last not least i thank the forschungszentrum karlsruhe , especially h. blmer , for permission to use its facilities even after retirement .
+ g. v. kulikov and g. b. khristiansen , soviet phys .
jetp * 35 * ( 1959 ) 441 j. n. stamenov et al . , trudy fian sssr * 109 * ( 1972 ) 132 r. glasstetter ( kascade collaboration ) , proc .
@xmath55 internat .
cosmic ray conf .
durban 1997 , contrib .
he 2.1.5 ; h. yoshii , j. phys .
japan * 32 * ( 1972 ) 295 j. r. hrandel et al .
( kascade collaboration ) , proc .
@xmath56 internat .
cosmic ray conf .
, salt lake city 1999 , contrib .
he 2.2.41 a. haungs et al .
( kascade collaboration ) , proc .
@xmath56 internat . cosmic ray conf . ,
salt lake city 1999 , contrib .
he 2.2.02 ; t. antoni et al .
( kascade collaboration ) , _ muon density measurements with the kascade central detector _ , astropart .
, in press b. peters , nuovo cim .
* 22 * ( 1961 ) 800 b. wiebel - sooth , p. l. biermann and h. meyer , astron
. astrophys . * 330 * ( 1998 ) 389 s. i. nikolsky and v. a. romachin , phys . atom .
* 63 * ( 2000 ) 1799 k. h. kampert et al .
( kascade collaboration ) , proc .
@xmath56 internat .
cosmic ray conf .
, salt lake city 1999 , contrib .
og 1.2.11 a. d. erlykin and a. w. wolfendale , astropart .
* 7 * ( 1997 ) 1 , 203 , * 10 * ( 1999 ) 69 ; j. phys .
* 23 * ( 1997 ) 979 , * 26 * ( 2000 ) 203 ; astron .
* 350 * ( 1999 ) l1 , * 356 * ( 2000 ) l63 m. aglietta et al .
( eas - top collaboration ) , astropart .
* 10 * ( 1999 ) 1 g. navarra , private communication ( 2000 ) m. a. k. glasmacher , phd thesis , the university of michigan ( 1998 ) m. a. k. glasmacher et al .
( casa - mia collaboration ) , astropart . phys . * 10 * ( 1999 ) 291 m. nagano et al .
, j. phys .
g : nucl . part .
* 10 * ( 1984 ) 1295 ; m. nagano , private communication ( 2000 ) m. nagano et al .
, j. phys .
g : nucl . part .
18 * ( 1992 ) 423 g. heinzelmann , private communication ( 2000 ) r. glasstetter , private communication ( 2000 ) a. chilingarian , private communication ( 2000 ) ; a. chilingarian et al .
@xmath56 internat .
cosmic ray conf .
, salt lake city 1999 , contrib .
he 2.2.9 n. n. kalmykov , private communication ( 2000 ) h. o. klages et al .
( kascade collaboration ) , nucl .
b ( proc . suppl . ) * 52b * ( 1997 ) 92 j. h. weber , phd thesis , university of karlsruhe ( 1999 ) ; report fzka 6339 , forschungszentrum karlsruhe ( 1999 ) ( in german ) s. v. ter - antonyan and l. s. haroyan , hep - ex/0003006 j. knapp , d. heck and g. schatz , report fzka 5828 , forschungszentrum karlsruhe ( 1996 )
|
28 size spectra of extensive air showers in the knee region from 7 different experiment are analysed consistently .
they are fitted by adjusting either 4 or 5 parameters : knee position , power law exponents above and below the knee region , overall intensity and , in addition , a parameter describing the smoothness of the bend .
the residuals are then normalized to the same knee position and averaged .
when 5 parameters are employed no systematic deviation from a simple smooth knee is apparent at the 1 % level up to about a factor of 4 above the knee . at larger shower sizes
a moderately significant deviation can be seen whose shape and position are compatible with a second knee caused by iron group nuclei .
cosmic rays ; knee ; eas
|
recently a very interesting stochastic model with deposition and evaporation processes has been introduced by barma _ et .
_ @xcite . in this model one
deposits atoms on to @xmath4 adjacent vacant sites of a @xmath5-dimensional lattice and evaporates atoms from any @xmath4 adjacent occupied sites , with specific rates for deposition and evaporation .
the cases @xmath6 and @xmath7 are exactly solvable , the former being trivial , and the latter being equivalent to ferromagnetic heisenberg spin-@xmath8 chain . for @xmath9 , on a linear chain of length @xmath10 , the phase space of this @xmath4-mer model consisting of the @xmath11 possible configurations is found to break up into an exponentially large number of dynamically disconnected sectors @xcite . this may be understood as being due to the existence of an infinite number of independent conserved quantities in this model @xcite .
these conservation laws also give rise to a wide range of relaxation behavior @xcite .
for example the density - density autocorrelation function shows different power law decays in different sectors ( @xmath12 ) and even a stretched exponential decay in some sectors . for @xmath9 ,
though there exist an infinite number of constants of motion , these still have not enabled us to get a full solution of the model so far . in this case
the number of sectors and their sizes in the steady state has been calculated exactly , but the dynamics is understood only qualitatively .
the quantum hamiltonian corresponding to the stochastic matrix has @xmath4-body interaction terms which makes the problem difficult to tackle analytically .
thus it is worthwhile studying models which show the same qualitative features and are at the same time more tractable analytically . with this motivation
we define a variant of the @xmath4-mer deposition - evaporation model with @xmath13 states per site .
this model for @xmath14 corresponds to the dimer deposition evaporation model studied by barma _
et.al_. for @xmath15 this model shares many qualitative features with the trimer model , like the existence of an infinity of conservation laws encoded as the conservation law of the `` irreducible string '' @xcite , and different power law decay of the autocorrelation function in different sectors .
however our model has the two advantages : it has only @xmath16-body interaction and it has an additional discrete symmetry group of very large order .
these properties make this model more tractable analytically than the @xmath4-mer model .
in fact , one can get many exact eigenvectors of the stochastic matrix in our model very easily .
however , we have not succeeded so far in solving this model completely . for simplicity
we study only @xmath17 case in this paper , for higher @xmath13 results will be qualitatively the same .
this paper is organized as follows . in section
@xmath16 we define our model and write the stochastic evolution operator in the form of the hamiltonian of a quantum many - body system of @xmath18 species of particles . in section
@xmath18 we show that the vector space consisting of @xmath19 configurations breaks up into an exponentially large number of dynamically disconnected sectors , and also calculate their sizes . in section
@xmath20 we describe the additional symmetries of this model and study the resulting further decomposition of the phase space formed by all the configurations belonging to a sector into subspaces .
we construct a class of exact eigenvectors in the next section . in section
@xmath21 we study the decay of autocorrelation function in various sectors . as in the case of trimer model @xcite
, we find sector dependent decay .
the spatial correlation function in the steady state for the sector corresponding to an initial condition of all sites in the same state is calculated in section @xmath22 .
it has power law decay with exponent @xmath2 .
a numerical diagonalization study of the stochastic matrix to find the dynamical exponent in this sector is presented in section @xmath23 .
our model is defined as follows : at each site of a one dimensional lattice , there is a spin variable @xmath24 which can be in any of the three states @xmath25 or @xmath26 .
these spin variables undergo a stochastic time evolution given by the following rule : any pair of adjacent spins , which are in the same state , can flip together to any of the other two states with some specified rates for the transitions . for example
an @xmath27 pair can flip to become either a @xmath28 pair or a @xmath29 pair .
this process can be thought of as the evaporation of a @xmath27 dimer and immediate deposition of a @xmath28 or @xmath29 dimer at that place .
we shall call this model the dimer deposition - evaporation model ( dde model ) .
this is a special case of the general reaction - diffusion process ( @xmath30 ) recently studied by dahmen @xcite .
we consider only the case of equal rates for all the transitions in this paper .
any configuration @xmath31 on the lattice can be represented by the basis vector @xmath32 , in a vector space of dimension @xmath19 .
if @xmath33 denote the probability of the configuration @xmath31 at time @xmath1 , then the master equation describing the evolution of these probabilities can be written as = w + where @xmath34 and @xmath35 is the stochastic matrix .
we can write @xmath35 in the form of a quantum hamiltonian as follows .
consider a quantum mechanical system of 3 species of particles on a one dimensional lattice of length @xmath10 with periodic boundary condition .
each species is characterized by a color @xmath13 which can be either @xmath25 or @xmath26 .
particles of the same colour experience a hard core interaction so that not more than one particle of the same color can occupy a given site .
therefor every site can be either empty or occupied by up to @xmath18 particles of different colors .
thus there are @xmath23 possible states on every site .
let @xmath36 and @xmath37 denote the pauli operators which annihilate and create a particle of color @xmath13 at the site @xmath38 . now consider the hamiltonian h = _
i=1^l a_i , i+1 + b_i , i+1 + c_i , i+1 [ hami ] where @xmath39 where the operator @xmath40 acting on any state results a nonnull vector only when @xmath41 and is given by , a_i , i+1 = + - 2 [ action ] the operators @xmath42 and @xmath43 have similar actions .
the operator n_i
= a_i^+a_i + b_i^+b_i + c_i^+c_i counts the number of particles at site @xmath38 . clearly = 0 , and the number of particles is conserved at each site .
if we restrict ourselves to the subspace of the full hilbert space in which there is only one particle at every site , _ ie .
_ @xmath44 for all @xmath38 , corresponding to every configuration @xmath45 in this subspace there is a unique configuration specified by the same l - string in the dimer problem .
the action of @xmath46 on @xmath47 is the same as that of @xmath35 .
thus @xmath35 is represented by the quantum hamiltonian given by ( [ hami ] ) acting on the subspace with @xmath44 for all @xmath38 .
make the particle - hole transformation ( @xmath48 ) on all odd numbered sites for all @xmath13 .
we define the current operator by j_i , i+1 = a_i^+ a_i+1 + b_i^+ b_i+1 + c_i^+ c_i+1 .
then the hamiltonian up to the addition of a constant can be written as h = _
i j_2i,2i+1 j_2i,2i+1^+ + j_2i,2i-1 j_2i,2i-1^+ + 3 _ i , n_i , n_i+1 , .
the dimer model now correspond to @xmath46 operating on a sector where @xmath44 for all even numbered sites and @xmath49 for all odd numbered sites .
the corresponding stochastic process is one in which there is a constant rate of exchange of any two particles at nearest neighbour sites .
as the original trimer model , this model has an infinite number of independent constants of motion . these are described most simply in terms of the `` irreducible string''(is ) @xcite .
we consider free boundary conditions for convenience .
each configuration can be represented as a string of @xmath10 characters , each character being one of @xmath25 and @xmath26 . from the string corresponding to a given configuration c ,
delete all adjacent pairs of @xmath27 , @xmath28 and @xmath29 .
each such deletion decreases the length of the string by @xmath16 .
repeat this operation on the resulting string until no more deletions are possible .
this defines the is corresponding to c. it is straightforward to extend the arguments of @xcite to our model , and show that the is is a constant of motion .
in addition , any two configurations having the same is can be reached from each other . thus the is can be used to label the different sectors into which the vector space breaks up . consider the set of all configurations in which no adjacent pair of sites are in the same state .
these configurations will not evolve under the given dynamics and are said to be fully jammed .
for these configurations the length @xmath50 of the irreducible string equals @xmath10 . the total number of such configurations is easily seen to be equal to @xmath51 .
now consider sectors which are not totally jammed ( @xmath52 ) .
the number of sectors labelled by irreducible strings of length @xmath50 will be the number of distinct irreducible strings of length @xmath50 which is @xmath53 .
the total number of sectors can be easily seen to be @xmath54 .
now we proceed to calculate the sizes of these sectors , _
i.e. _ the number of distinct configurations which belong to a particular sector . the size of a fully jammed sector is clearly @xmath55 .
this is because a totally jammed configuration can not evolve and hence is the only member of that sector .
let @xmath56 be the size of the sector labelled by irreducible string @xmath57 on a lattice of length @xmath10 .
we define the generating function g(is , z ) = _ l = l^ d(is , l)z^l where the summation over @xmath10 extends from @xmath50 , the length of @xmath57 to infinity .
of special interest is the sector corresponding to is of length zero .
this we will call the null sector and denote its irreducible string by @xmath58 . to compute @xmath59 , we introduce a formal series which is a sum of all strings of arbitrary length , which are made up of three letters @xmath60 , @xmath61 and @xmath26 , and reducible to @xmath58 .
a string @xmath62 is said to be _
, if the irreducible string @xmath63 and it can be written as @xmath64 such that @xmath65 . if @xmath66 denote the sum of all indecomposable strings which are reducible to @xmath58 then @xmath67 satisfies the equation g()= [ eq : g1 ] further if @xmath68 denotes the the sum of all indecomposable strings which are reducible to @xmath58 and starting with the letter @xmath13 then g_i()=g_i^a+g_i^b+g_i^c [ eq : g2 ] @xmath68 is given by g_i^q = q . .
q [ eq : g2.1 ] to get the generating function we replace all occurences of @xmath69 by @xmath70 in equation ( [ eq : g2.1 ] ) .
then @xmath71 and @xmath72 all reduce to the same power series in @xmath70 , call it @xmath73 , given by g(z ) = this equation determine @xmath73 as an explicit function of @xmath70 given by g(z ) = ( 1 - ) /4 .
[ eq : g ] using equations ( [ eq : g1 ] ) and ( [ eq : g2 ] ) , we get g(,z)= the growth of @xmath74 for large @xmath10 is determined by the singularities of @xmath73 nearest to the origin .
this happens at @xmath75 . and
since @xmath76 has a square root singularity near @xmath77 , it is easily verified that d(,l)~8^l/2 l^-3/2 for large @xmath10 .
one can easily generalize this procedure to find the size of the sector when the irreducible string is of finite length @xmath50 .
let @xmath78 , where @xmath79 or @xmath26
. then @xmath80 ( defined similarly as @xmath67 ) is given by g(is)=._1 .. _2. .._l.g ( ) where @xmath81 .
replacing @xmath69 by @xmath70 as before we get g(is , z)=[]^l [ eq : g1 ] where @xmath73 is given by equation ( [ eq : g ] ) and @xmath50 is the length of the irreducible string @xmath57 .
then the size of the sector characterised by the irreducible string is is the coefficient of @xmath82 in ( [ eq : g1 ] ) , and hence d(is , l)=_c dz where integration is over a small circle encircling the origin . from this
it is straightforward to find the asymptotic behavior of @xmath56 by doing the integral using the saddle point approximation .
let @xmath83 . then d(is , l)~[k()]^l , [ sec_size1 ] where k()=(1-)^-(1-)/2 ( 1+)^-(1+)/2 2^(3-)/2 .
[ sec_size2 ] thus , in general the size of a sector increases exponentially with @xmath10 and the growth constant depends on the density of the irreducible string . from this one
can recover the results for the null and totally jammed sectors by noting that @xmath84 and @xmath85 .
in the previous section we discussed the decomposition of the @xmath19 dimensional configuration space of the model into @xmath54 mutually disjoined sectors . this sector decomposition described in terms of the irreducible string @xmath57
is maximal in the sense that it is possible to go from any configuration to any other configuration in the same sector .
however , this only helps us to block diagonalize @xmath35 . in order to do better
, we have to find additional operators that commute with @xmath35 .
these operators are not completely diagonal in the configuration basis .
we now describe how to construct such operators using the additional symmetries of our model .
these symmetries are not present in the original @xmath4-mer model .
they are also not present in our model if all the transition rates are not equal .
using these symmetries we are able to reduce the task of diagonalizing the stochastic matrix with in a sector , by further block diagonalizing it .
the deposition - evaporation model can be recast in the form of a generalized interface growth model , similar to the kpz model , where the variable at each site is a @xmath86 matrix instead of a scalar .
this is seen as follows : we consider three non commuting matrices @xmath87 and @xmath88 such that @xmath89 , the identity matrix .
a simple choice which satisfies these conditions is the unimodular matrices rclrclrcl a(a)&= & ( cc1&0 + 1&-1 ) , & a(b)&= & ( cc1&1 + 0&-1 ) , a(c)&= & ( cc1&0 + 0&-1 ) alternately , we could choose them to be @xmath90 unitary matrices .
for a configuration @xmath91 , we associate a @xmath86 matrix @xmath92 to every site @xmath38 given by i_i(c ) = a(q_i) a(q_2)a(q_1 ) .
it is easy to see that the stochastic evolution of the dimer model can be cast in terms of matrices @xmath93 s by the following local evolution rule : if @xmath94 , then @xmath95 is reset randomly to any one of the three values @xmath96 at a constant rate say @xmath55 .
the conservation law of the @xmath57 in this representation becomes the simple statement that @xmath97 is unchanged .
a configuration of dimers is then equivalently characterized by specifying the matrix variables @xmath95 at each site @xmath38 .
this matrix formulation brings out clearly the existence of a discrete group of infinite order in the model .
if we work with free boundary conditions the set of allowed values of the of the matrices @xmath93 s can be put in one to one correspondence with the sites of a @xmath18-coordinated bethe lattice .
identify one site of the bethe lattice as the origin and associate the identity matrix to this site .
we color the bonds of this bethe lattice by three colors @xmath69 such that any three bonds meeting at a site are all of different colors .
the matrix @xmath98 corresponding to a site @xmath99 of the bethe lattice is given by the ordered product of @xmath100s @xmath101 along the unique path from the origin to @xmath99 .
a sequence @xmath102 , @xmath103 to @xmath10 specifies a configuration of the dde model , and also a unique @xmath10-step path on the bethe lattice which starts at the origin . as an example , the path corresponding to the configuration @xmath104 is shown in figure [ b_l_rep ] .
configurations in the null sector correspond to paths that return to the origin .
any pair of adjacent spins that are in the same state in the dimer configuration will corresponds to an immediate retraversal of a step on the bethe lattice .
the stochastic evolution in the dde model leads to a stochastic evolution of this @xmath10-step polymer chain on the bethe lattice
. an elementary movement of this polymer chain dynamics is illustrated in figure [ b_l_rep ] , where the transition from the configuration @xmath105 to @xmath106 is shown .
the operation of interchange of one branch of the bethe lattice with another is a symmetry of @xmath35 since this only corresponds to a recoloring of the bonds .
this is illustrated in figure [ br_flip ] , where we have considered only one branch starting from the origin .
at every site @xmath99 on the bethe lattice define an operator @xmath107 which interchanges the two subtrees starting from that site and going away from the origin .
@xmath108 changes a configuration @xmath109 to @xmath110 where @xmath111 , where @xmath112 is a matrix valued function of its argument .
clearly @xmath113=0 $ ] .
we say that site @xmath114 is descendant of site @xmath99 if the unique path connecting @xmath114 to the origin goes through @xmath99 . from the figure
it can be seen that @xmath115=0 $ ] if neither @xmath99 nor @xmath114 is a descendant of the other . if @xmath114 is a descendant of @xmath99 then @xmath116 , where @xmath117 is the operator at site @xmath118 which is obtained from site @xmath114 by the action of @xmath107 . in figure [ br_flip ] for example , @xmath119=0 $ ] . but
@xmath120\ne 0 $ ] since @xmath16 is a descendant of @xmath55 .
instead they satisfy the relation @xmath121 .
this symmetry comes from the geometrical symmetry of the underlying bethe lattice .
this in turn is related to the internal color symmetry in the model , and we will call it recoloring symmetry c. in addition to these one has a permutation symmetry of order @xmath18 ( @xmath122 ) about the origin which corresponds to interchanging the @xmath18 color labels globally . the set of operators @xmath123 together with the operators corresponding to @xmath122 will constitute an infinite set of conserved quantities , which are however not all commuting with each other .
one interesting consequence of the recoloring symmetry is that the spectrum of @xmath35 in a sector depends only on the length @xmath50 of the irreducible string and not on its details .
this can be seen as follows : for a general sector , the endpoints of all the paths on the bethe lattice will be fixed .
the shortest path connecting these two points corresponds to the irreducible string .
all @xmath124s such that site @xmath99 does not belong to this path leaves the irreducible string and hence the sector it labels invariant . on the other hand one can go from one irreducible string to any another one having the same length by a series of branch flip operations using the @xmath124 operators which are on the shortest path . since each of these operations are symmetry operations of @xmath35 , _
i.e. _ @xmath125=0 $ ] for all @xmath99 , the spectrum will be identical for all these sectors .
this property is not present in the original @xmath4-mer model .
we can make use of these symmetries for further block diagonalisation of @xmath35 with in a sector .
this is achieved by breaking the state space spanned by all the configurations in a particular sector into subspaces .
we illustrate this with a few examples . for convenience we will consider the case of null sector .
first consider a lattice of length @xmath16 .
there are only three configurations in this sector which are @xmath126 and @xmath127 . in this case
we can get all the eigenstates of the hamiltonian using @xmath122 .
consider the states lcl & = & + + + & = & + + ^2 + & = & + ^2 + .
where @xmath128 is the cube root of unity .
these are the eigenstates of the operator corresponding to cyclic permutation of @xmath60 , @xmath61 and @xmath26 ( @xmath129 ) with eigenvalues @xmath130 and @xmath131 respectively . since @xmath132=0 $ ] these are also the eigenstates of @xmath35 .
the eigenvalues of @xmath35 corresponding to these eigenvectors are @xmath133 .
thus in this case we were able to diagonalize @xmath35 completely .
but this is a rather trivial example .
next we consider the case of @xmath134 .
there are @xmath135 states in this sector .
first we consider the symmetry operations about the sites @xmath55 , @xmath16 and @xmath18 as shown in figure [ bl_l4 ]
. consider the states lcl & = & ( - ) + & = & ( - ) + & = & ( - ) .
+ [ psi ] these are the non - degenerate eigenstates with eigenvalue @xmath136 of @xmath137 , @xmath138 and @xmath139 respectively .
it can be verified very easily that these are also eigenstates of @xmath35 .
this gives us a decomposition of the @xmath135 dimensional vector space into subspaces of dimensions @xmath140 and @xmath141 .
using the cyclic permutation symmetry at the origin , we can get a further decomposition of the @xmath141 dimensional subspace .
for this consider the eigenstates of @xmath129 lcl & = & ( + + ) + & = & ( + + ) + & = & ( + + ) + & = & ( + + + & & + + + ) + & & + & = & ( + + ^2 ) + & = & ( + + ^2 ) + & = & ( + + ^2 ) + & = & ( ( + ) + ( + & & + ) + ^2(+ ) ) + & & + & = & ( + ^2 + ) + & = & ( + ^2 + ) + & = & ( + ^2 + ) + & = & ( ( + )
+ ^2 ( + & & + ) + ( + ) ) + [ phi ] since @xmath129 commutes with @xmath35 , @xmath142 , @xmath143 and @xmath144 form @xmath18 invariant subspaces under the action of @xmath35 .
thus we have got the following decomposition of the @xmath135 dimensional vector space spanned by all the configurations in the null sector in to subspaces of dimensions . 1 +
1 + 1 + 4 + 4 + 4=15 we have carried out the same procedure for the case of @xmath145 . in this case
the size of the null sector is @xmath146 . using the @xmath124 operators , this
can be broken into 18 subspaces out of which @xmath141 are of size @xmath55 , @xmath18 are of size @xmath22 and another @xmath18 are of size @xmath147 .
a further decomposition of these subspaces can be achieved using the symmetry of dynamics under reflection . using both the color and reflection symmetries ,
the resulting form of the decomposition of the vector space for the case @xmath145 is 87 = 121 + 33 + 34 + 36 + 3 12 . where @xmath148 means @xmath149 subspaces of size @xmath150 . in principle
this procedure can be carried out for a lattice of any length . for a linear chain of length @xmath10 , sites up to distance @xmath151 from the origin on the bethe lattice are reachable .
the number of sites that are one step away from the boundary and inside will be @xmath152 ( @xmath153 ) . as we have seen for @xmath134 , the antisymmetric eigenstates of the @xmath154 operators on each of these sites will be an eigenstate of @xmath35 .
thus one can easily get a large number of eigenstates in this model . by constructing the antisymmetric eigenstates of the @xmath154 operators on sites which are interior to this
, one can get invariant subspaces as we have seen in the case of @xmath145 ( the 3 sectors of size 7 ) .
so far we have not been able to get a general result for the total number of subspaces and their sizes within a sector as a result of this symmetry .
but one can use this symmetry for the numerical diagonalization of the stochastic matrix for small sizes , where one can restrict oneself to one subspace ( say the symmetric subspace where all the eigenvalues of the @xmath155 operators are @xmath55 ) .
this we have done to estimate the dynamical exponent in the null sector , where we have seen that these additional symmetries lead to substantial decrease in the size of the matrices to be diagonalized .
these results are discussed in section @xmath23
in the previous section we have constructed some of the eigenvectors of the stochastic matrix in the null sector using the recoloring symmetry of the dynamics of @xmath10-step chain on the bethe lattice . in this section
we show that the same procedure can be used to construct eigenvectors in sectors which are almost totally jammed , _
i.e. _ the length of the irreducible string @xmath156 , where @xmath150 is a small integer . as we explained in the earlier section ,
the spectrum of @xmath35 is the same for all the sectors whose irreducible strings are of same length .
so we need to consider only one representative sector which we choose to be the one in which the irreducible string is @xmath157 .
consider the case when @xmath158 .
we use periodic boundary condition for convenience .
there are two types of configurations in this sector . in the first type
there will be three adjacent sites , say @xmath159 and @xmath160 which are of the same color ( either @xmath161 or @xmath162 _ e.g. _ @xmath163 ) .
let these states be denoted by @xmath164 . in the second type there will be two adjacent sites , say @xmath165 and @xmath160 which are having color @xmath26 .
( _ e.g. _ @xmath166 .
let this type of states be denoted by @xmath167 .
in both these types site @xmath165 and site @xmath160 have the same color and thus constitute a reducible part
. it can be very easily seen that under the deposition - evaporation dynamics , the position of the reducible part makes a nearest neighbour random walk .
hence the dynamics in this sector is equivalent to a random walk problem with two states per site ( @xmath168 and @xmath169 ) .
the eigenfunctions @xmath170 and eigenvalues @xmath171 of the stochastic matrix can be found by solving this random walk problem .
let @xmath172 and @xmath173 be defined by @xmath174 and @xmath175 .
they satisfy the following eigenvalue equations lcl _ k _ k(i)&=&_k(i-1 ) + _ k ( i+1 ) + _ k(i-1 ) + _ k(i ) -4 _ k(i ) + & & + _ k _ k(i)&=&_k(i ) + _ k(i+1 ) - 2 _ k(i ) the solution for this eigenvalue equations are given by _
k=(3-(k ) ) and lcl _ k(n)&=&e^ikn + _ k(n)&=&(e^ikn(1+e^ik ) ) + where @xmath176 . in the case of free boundary conditions
some of the localized eigenvectors can be constructed directly with out solving this random walk problem .
consider the situation where the reducible part is near one of the boundaries .
_ i.e. _ the states @xmath177 and @xmath178 .
it is easy to see that the state obtained by antisymmetrizing , @xmath179 , is an eigenstate with eigenvalue @xmath136 .
one more eigenstate can be constructed by antisymmetrizing about the other end .
now consider the case of @xmath180 . in this case
the reducible part is of length @xmath20 .
a typical configuration in this sector will be @xmath181 .
the reducible parts @xmath182 and @xmath183 perform random walk under the dynamics .
when these random walkers come closer to each other they can form state like @xmath184 or an intermediate state like @xmath185 .
there are two such intermediate states which corresponds to the two values @xmath186 .
the two random walkers can go into such intermediate states and remain there for a while by making transition between them .
this can be considered as a short range attractive interaction between the two walkers .
so the dynamics in this sector is equivalent to a problem of two random walkers on a line having a short range attractive interaction .
we can use the same procedure of antisymmetrization to get some more the eigenvectors .
one type of eigenvector is obtained by antisymmetrizing about both ends simultaneously .
= - the second type of eigenvector is obtained by antisymmetrizing two intermediate states @xmath187 and @xmath188 .
= - since there are @xmath10 positions on the string , where @xmath189 can occur , for the case of periodic boundary conditions , there are @xmath10 such eigenvectors .
this can be easily generalized for the case where there are @xmath190 random walkers ( @xmath191 ) .
a typical intermediate state formed by pair wise union of random walkers will have @xmath150 number of substrings of the form @xmath192 .
the eigenvector is constructed by antisymmetrization of all these .
the corresponding eigenvalue will be @xmath193 .
since these @xmath150 substrings can be arranged in roughly @xmath194 ways , we can get so many eigenvectors . as an example , in the case of @xmath195 one such eigenvector is given by ccc & = & - + & & + - +
for the trimer model , the existence of the infinite number of conservation laws lead to non - exponential relaxation in equilibrium @xcite . it is found that the density - density autocorrelation function decay as a power law for large times and the exponent is different in different sectors . in most sectors
the decay is @xmath196 but is of the form @xmath197 in most of the sectors with periodic irreducible strings . in some special sectors with periodic irreducible string ,
stretched exponential behavior ( exp(-@xmath198 ) ) is found . in the null sector
the autocorrelation function is found to decay as @xmath199 , where @xmath200 .
this diversity of relaxational behaviour has been explained in terms of the hard core random walkers with conserved spins ( hcrwcs ) model @xcite .
it has been argued on qualitative grounds that these two models are in the same universality class .
however , a strict proof of this proposition is not yet available . an important assumption of the hcrwcs model is that the spin carried by each walker is conserved , but does not affect the diffusive motion of the walkers in any way .
this is not strictly true in the trimer model .
however , for our model , this property can be exactly established as a consequence of the recoloring symmetry .
thus the argument which predict the different behaviour of spin - spin autocorrelation function in different sectors is much cleaner for this model than for the trimer model studied earlier .
we proceed to give the arguments which are an adaptation of the arguments for the trimer deposition evaporation model @xcite , in some detail . the time dependent spin - spin autocorrelation function in the steady state is defined by c_^i(t ) = - . where prob(@xmath201 ) denote the probability of event @xmath201 and @xmath202 is the state of the spin at the @xmath38th site at time @xmath1 .
each of @xmath99 and @xmath114 is either @xmath25 or @xmath26 . since @xmath99 and
@xmath114 each takes @xmath18 values , one can define @xmath203 autocorrelation functions in this model . among these , by time reversal symmetry @xmath204 and also @xmath205 .
thus there are only @xmath18 independent correlation functions . as in the trimer model
this autocorrelation functions shows sector dependent decay @xcite .
let us consider the following five different representative sectors characterised by the irreducible strings @xmath206 .
@xmath207 , @xmath208 @xmath209 , @xmath210 .
@xmath211 , @xmath212 . random string and @xmath213 .
@xmath58 ( null sector ) , where @xmath214 denotes repetition .
the length of the irreducible string in the first four cases is taken to be @xmath151 .
we will first present a qualitative theory for the decay of autocorrelation functions and then compare its predictions with our monte carlo data .
the behavior of the autocorrelation function can be understood in terms of the random walk of the characters which constitute the irreducible string .
consider a sector labelled by an irreducible string @xmath215 , with the length of the irreducible string @xmath50 is a finite fraction of the total length @xmath10 .
let @xmath216 be the positions of the characters in a given configuration @xmath31 which do not get deleted under the deletion algorithm of section @xmath18 .
then @xmath217 ( @xmath103 to @xmath50 ) at all times .
we can think of @xmath216 as the positions of @xmath50 interacting random walkers with the @xmath150th random walker from the left carrying a color @xmath218 with it .
the positions of random walkers , @xmath219 will change in time as @xmath31 changes .
the walkers move either to the left or right but always remain on the same sublattice .
further , they do not cross each other .
we have argued in section @xmath20 that the spectrum of the stochastic matrix in this model is completely independent of the detail of the irreducible string string sequence @xmath220 and only depend up on the length of the sequence .
however , the autocorrelation involves a weighted sum of correlations of @xmath220 , which gives rise to different relaxational behavior in different sectors .
the important contribution to the autocorrelation function comes from times when the site @xmath38 is occupied by character from the irreducible string .
we shall assume with out proof that the contribution coming from times when the site @xmath38 is occupied by a reducible character is qualitatively similar .
since the hard core random walkers will remain on the same sublattice they were initially , we can perform a sublattice average of @xmath221 .
let @xmath222 denotes the sublattice which is either @xmath223 ( odd sites ) or @xmath224 ( even sites ) . in hcrwcs approximation we can write the expression for the autocorrelation function as c_^(t)_k=-^+p(k , t)m^(k ) [ c_ab_2 ] where @xmath225 gives the joint probability of finding two particles at a given site , one at time @xmath226 and the other at time @xmath1 such that the difference between their labels is @xmath227 .
@xmath228 is a measure of the correlation of the colors in the irreducible string , and is given by m^(k)=_m ( q_x_m , ) ( q_x_m+2k , ) , where the summation is over all the labels of the particles which belong to the sublattice @xmath222 . the probability @xmath225 can be obtained using the results of random walks of hard core particles @xcite . note that @xmath225 depends only on the total length of the irreducible string and not on its details . while @xmath229 depends only on the irreducible string and is independent of time .
this separation of the sector dependence and time dependence is the crucial feature which allows calculation of @xmath230 in different cases .
_ this separation is exact due to the recoloring symmetry in our model , but only approximate in the original trimer model .
_ equation ( [ c_ab_2 ] ) can be used to find the behaviour of autocorrelation function .
for example consider the sector labelled by the irreducible string @xmath157 . on the @xmath223 sublattice @xmath231 for all @xmath149 and hence c_aa^a(t)=_k=-^+p(k , t )
this is the density - density correlation function of hard core particles which decays as @xmath197 for large @xmath1 @xcite .
our simulation gives the same power law decay which is shown in figure [ abab_aa ] . from the symmetry between @xmath61 and @xmath26 on this sublattice and the constraint @xmath232
, it can be easily seen that @xmath233 for all values of @xmath99 and @xmath114 .
the behavior of the correlation functions on both the sublattices is the same which follows from the symmetry of the irreducible string . in the case of the sector labelled by the irreducible string @xmath234
, @xmath235 is of the form @xmath236 , where @xmath237 and @xmath238 are constants independent of @xmath4 .
thus @xmath239 has two parts . the first part decay like @xmath197 as explained earlier .
the second part is the fourier transform of @xmath225 which goes as exp ( -@xmath198 ) for large @xmath1 .
the stretched exponential decay dominates the short term behavior but asymptotically the behavior will cross over to @xmath197 . by symmetry @xmath240 should have the same decay . on the @xmath224 sublattice @xmath241
is @xmath26 for all @xmath149 , so the decay will be purely diffusive ( @xmath197 ) as explained earlier .
these behaviours can be seen in figure [ acbc_aa ] . for the sector @xmath211 , the decay of @xmath239
is shown if figure [ abc_aa ] .
the leading behavior of the correlation function can be shown to be stretched exponential decay by a similar argument as in the previous case . by symmetry of the irreducible string
the behaviour should be the same for all the correlation functions and on both the sublattices .
this has been confirmed in our simulation . when the irreducible string is random , @xmath228 is significant only for small values of @xmath4 . for small @xmath4 , @xmath242 for large @xmath1 @xcite .
therefore the autocorrelation has a @xmath196 decay which can be seen in figure [ null_aa ] .
figure [ null_aa ] also shows the decay of @xmath239 in the null sector .
it is a power law decay with the value of the exponent @xmath243 .
the same power law decay found for other values of @xmath99 and @xmath114 and on both sublattices , which follows from symmetry .
barma _ et.al .
_ have obtained same estimate for this exponent for the trimer model in the null sector @xcite .
this is an evidence that the dynamics in these two models for the case of the initial condition in which all sites are in the same state , belongs to the same universality class .
we can understand the behaviour of the correlation functions in sectors where the irreducible string has a finite density using the hcrwcs model , this approach is no longer useful when studying the null sector .
an analytical understanding of the temporal decay of autocorrelation functions in this sector is still lacking .
let @xmath244 be the probability that , in the steady state , spins at two sites which are separated by a distance @xmath245 are @xmath99 and @xmath114 .
we calculate this equal time spatial correlation function for the null sector .
let us assume periodic boundary conditions .
let @xmath62 be the string corresponding to a given configuration on this lattice with sites @xmath38 and @xmath246 are occupied by @xmath99 and @xmath114 .
let @xmath247 be the substring of length @xmath248 formed by spins between @xmath99 and @xmath114 and @xmath249 be the substring of length @xmath250 formed by the remaining spins .
the endpoints of the substring @xmath249 are @xmath99 and @xmath114 .
let @xmath251 and @xmath252 be the irreducible strings corresponding to @xmath247 and @xmath249 . since @xmath63 , @xmath252 will be @xmath251 reflected about one end .
we may write @xmath253 .
c_,(r)= [ corr1 ] where the summation is over all the distinct irreducible strings of @xmath247 .
@xmath254 is the size of the sector labeled by @xmath57 on a lattice of length @xmath248 with free boundary conditions and @xmath255 is the size of the sector labeled by @xmath256 on a lattice of length @xmath250 with boundary sites fixed at @xmath99 and @xmath114 .
though we can calculate exact expressions for each of the above quantities , we derive the asymptotic behaviour by taking @xmath257 . the length of the irreducible string @xmath258 , therefor in this limit one can make the approximation @xmath259 . in the case of free boundary condition
the size of a sector depends only on the length of the irreducible string ( [ sec_size1],[sec_size2 ] ) and we may replace the sum over distinct irreducible strings by a sum over the lengths of the irreducible strings as follows .
c_,(r)= [ corr1.1 ] where @xmath260 is the number of distinct irreducible strings of length @xmath50 , which is given by @xmath261 . for large @xmath10 ,
hence in the summation over @xmath50 only terms corresponding to small @xmath50 will contribute .
we can find the asymptotic @xmath245 dependence by approximating the sum by only the @xmath263 term .
then we have c_,(r)~d(,r-1)d_,(,l - r+1)/d(,l ) .
for large @xmath10 gives c_,(r)~ for large r , where @xmath264 is a constant independent of @xmath245 but depends on @xmath99 and @xmath114 . from the symmetry between different colors
, it follows that @xmath265
as mentioned in the section @xmath21 , the decay of autocorrelation function in the null sector suggests that the dynamics in this sector belongs to a new universality class .
the description of dynamics in terms of the random walk of the characters of the irreducible string is not possible in this sector .
one way of studying this sector is to directly diagonalize the stochastic matrix for small sizes .
however this task is not very easy because the size of the matrix to be diagonalized grows very fast with the size of the lattice ( @xmath266 ) .
one can make use of various symmetries of the model to block diagonalize the matrix first and hence reduce the size of the matrix to be diagonalized . by using translation , reflection and the recoloring symmetry described in section @xmath20 we have been able to diagonalize the stochastic matrix for lattice sizes up to @xmath267 , in the fully symmetric subspace .
the algorithm used here for diagonalizing the matrix using the various symmetries is similar to the one we have used for studying trimer model @xcite . for details of the algorithm
we refer the reader to this paper . in table
@xmath55 , we have listed the size of the matrix ( @xmath268 ) obtained after using all the three symmetries and the value of the largest nonzero eigenvalue in the symmetric subspace in the null sector for lattice sizes @xmath10 from @xmath16 to @xmath267 . for comparison we also list the total size ( @xmath269 ) of the null sector , and also the reduced matrix size ( @xmath270 ) if only the recoloring symmetry c is taken into account .
we see that the additional recoloring symmetry gives a significant reduction in the size of the matrix to be diagonalized .
the largest eigenvalue of the stochastic matrix is @xmath226 .
the difference between the largest and the second largest eigenvalues gives the gap in the spectrum of the stochastic matrix .
we have computed this gap for lattice sizes from @xmath16 to @xmath267 . by using the finite size scaling relation @xmath271 , the dynamical exponent @xmath272 for a lattice of size @xmath10
is estimated from the relation z_l = ( ) / ( ) .
the estimate of @xmath273 for various values of @xmath10 are shown in table @xmath55 .
the convergent of @xmath273 seemed to be quite good , and the estimate of the extrapolated value of @xmath70 corresponding to @xmath274 is @xmath275 .
the error bar reflects our subjective estimate of the possible systematic errors in the extrapolation .
this value is in good agreement with the value of @xmath276 for the trimer model in the null sector @xcite , and is evidence that both these models are in the same universality class .
we thank m. barma for introducing us to the model , and for his continued interest and discussions throughout the course of this work .
we have benefited from several discussions with peter .
b. thomas in the earlier stages of this project .
|
we study a model of stochastic deposition - evaporation with recombination , of three species of dimers on a line .
this model is a generalization of the model recently introduced by barma _ et .
al . _
( 1993 _ phys .
rev .
lett . _ * 70 * 1033 ) to @xmath0 states per site .
it has an infinite number of constants of motion , in addition to the infinity of conservation laws of the original model which are encoded as the conservation of the irreducible string .
we determine the number of dynamically disconnected sectors and their sizes in this model exactly . using the additional symmetry
we construct a class of exact eigenvectors of the stochastic matrix .
the autocorrelation function decays with different powers of @xmath1 in different sectors .
we find that the spatial correlation function has an algebraic decay with exponent @xmath2 , in the sector corresponding to the initial state in which all sites are in the same state .
the dynamical exponent is nontrivial in this sector , and we estimate it numerically by exact diagonalization of the stochastic matrix for small sizes . we find that in this case @xmath3 .
# 1 pacs numbers : 02.50+s , 75.10j , 82.20.m , 05.50+q
|
the increasing demands for fast and reliable wireless communications have spurred development of multiple - antenna systems in order to efficiently harvest the capacity gains @xcite .
recent information - theoretic research indicates that a feedback channel can be further employed to furnish channel state information ( csi ) to the transmitter side , which may affect closed - loop capacity gains @xcite . with some form of knowledge of the wireless channel conditions
, the transmitter can adapt to the propagation conditions by the use of a variety of channel adaptive techniques @xcite . specifically , in a multiple mobile station ( ms ) scenario , with the knowledge of the channel to nearby co - channel mss , it is possible to actively suppress the signal to the interfered users and meanwhile maximize the effective signal power @xcite . in this case , the base station ( bs ) can obtain the required channel coefficients through a feedback channel from the ms .
then , mechanisms such as multiple - antenna precoding can be utilized to mitigate the effects of co - channel interference and exploit spatial dimensions to increase the capacity of wireless networks @xcite .
since csi is essential for closed - loop wireless communication systems , the techniques on how to effectively feedback csi from the transmitter to the receiver has been intensively studied @xcite . as perfect feedback of csi
is typically unavailable due to complexity or practicality constraints , the infinite feedback of csi is hard to realize in practice
. therefore , it is important to investigate how to control the amount of feedback signalling overhead according to the individual requirements in order to achieve good quality of service ( qos ) . as a result
, csi feedback rate control problem has attracted lots of attention in recent years @xcite . in @xcite , the quantized feedback approach for power - control is designed to minimize an upper bound of multiple - input - single - output ( miso ) systems .
recently , two specific forms of partial feedback , namely , channel mean feedback @xcite and channel covariance feedback @xcite , have been investigated for slow - varying and rapidly varying mimo channels , respectively .
the existing work typically treated each ms independently , and researched the multi - ms csi feedback problem in physical layer , e.g. , from either communication or information theory point of view .
this can not efficiently model the interactions among self - interested mobile users in wireless systems @xcite .
if the feedback channel is limited , there exists conflicts in effective csi feedback rates between each ms .
if one ms transmits too much csi , it will result in the reduction of the rest mss csi feedback amounts , and thus , degrade the others performance .
hence , it will be desirable to sort out the competition problem by finding a balance in this multi - ms feedback scenario , and meanwhile achieve better qos .
game theory @xcite offers a set of mathematical tools to study the complex interactions among interdependent rational players and predict their choices of strategies @xcite . in this paper , an alternative approach to the feedback rate control problem in wireless systems based on an economic model is proposed . in this model ,
each ms s preference is represented by a utility function , which quantifies the level of satisfaction a user gets from using the system resources @xcite .
each player in the game maximizes a utility function in a distributed fashion .
the game settles at a nash equilibrium if one exists .
since users act selfishly , the equilibrium point is not necessarily the best operating point from a social point of view @xcite . to achieve a more socially desirable result ,
a powerful tool by pricing the system resources can be introduced , which is able to guide user behavior toward a more efficient operating point @xcite . to the best of our knowledge ,
the game - theoretic methods are first applied to study csi feedback rate control under this economical model .
specifically , we investigate the scenario in a single - cell wireless data network , where a multiple - antenna bs communicates with a number of co - channel users through a minimum mean square error ( mmse ) precoder and each user tries to maximize its own utility .
two types of feedback protocols , fdma and csma , are investigated .
for ease of understanding , we first present a noncooperative feedback rate control game ( nfc ) , which optimizes individual utility in a distributed fashion . while the resulting noncooperative feedback rate control game has a nash equilibrium , it is inefficient from a social point of view
. therefore , we further introduce pricing to create cooperation between each ms in order to improve efficiency , called the noncooperative feedback - rate control game with price ( nfcp ) .
the price function is a linear function of the csi feedback rate that also allows a distributed implementation by broadcasting the price per bandwidth from the bs to all the mss .
it shows that there exists an equilibrium in the proposed nfcp .
simulation results indicate that by adjusting price , nfcp provides better overall utility than nfc .
in addition , the distributed nfcp game approach achieves near optimal performance compared to the centralized scheme , and thus , improves the overall throughput of wireless data networks .
the rest of the paper is organized as follows : in section ii , we introduce some preliminaries , including the system model , the multiple access protocols , and csi feedback rate model . in section iii , we describe the mmse precoder , and some properties by using the csi feedback rate model .
the proposed nfc and nfcp algorithms are described in iv .
simulation results are provided in section v. in section vi , we draw the main conclusions .
some derivations are given in the appendixes .
_ * notation * _ : boldface lower - case letters denote vectors , @xmath0 , @xmath1 and @xmath2 represent conjugate , transpose , and conjugate transpose , respectively , @xmath3 , and @xmath4 $ ] represents its variance .
in this section , we first give the system model . then
two types of feedback channels are discussed .
finally , the csi feedback rate model is illustrated . in this paper
, we consider a system in which a number of co - channel mss are served by one bs .
the architecture is depicted in fig.[fig : scenario ] .
the bs is assumed to know the linear processing performed by the mss , which can acquire the required csi through a feedback channel from the mss . using multiple antennas at the bs of a cellular system ,
transmit precoding can be performed for simultaneous transmission to several co - channel mobile users .
the precoder is designed assuming a stationary scenario in which the fast ( rayleigh ) fading is described by its second order properties .
we also assume narrow - band signals without any time dispersion , i.e. , the channel fading is frequency flat . for simplicity , we assume every ms is equipped with a single receive antenna .
we assume that the system works in a fdd model , where the bs has @xmath5 transmit antennas serving @xmath6 mss simultaneously in the same frequency band , while each ms feeds back the csi through different channels in order to better protect the control information by avoiding collisions . for the @xmath7-th ms , the input signal , @xmath8 ,
is first precoded by complex weights @xmath9 before transmitted from the @xmath5 antennas at the bs .
the corresponding output can be written as @xmath10 where @xmath11 .
the received signal at the @xmath7-th ms can be then expressed as @xmath12 where @xmath13^t\in\mathbb{c}^{n_t{\times}1}$ ] represents the channel coefficients from the bs to the @xmath7-th ms with zero mean and unit variance , @xmath14 is the awgn noise @xmath15 , @xmath16 is the desired signal , and @xmath17 can be treated as the interference .
note that the model can easily be extended to frequency selective channels , taking both co - channel interference and inter - symbol interference into account @xcite . in this subsection ,
two multiple access protocols for the uplink are presented . for simplicity , in the downlink
, the bs simultaneously serves multiple co - channel mobile users by performing precoding , while in the uplink two standard multiple access protocols are examined for csi feedback .
we assume that the total system bandwidth is @xmath18 , and the bandwidths for downlink and uplink transmissions are @xmath19 and @xmath20 , respectively .
then , we have @xmath21 the bs serves mss simultaneously in the same frequency band , while each ms feeds back the csi through orthogonal channels , i.e. , frequency division multiple access ( fdma ) , in order to better protect the control information by avoiding interference . recalling ( [ eq : rat ] ) , the uplink bandwidth can be then calculated as @xmath22 where @xmath23 denotes a scaling factor to transform the uplink csi feedback rate into bandwidth . and
the downlink bandwidth can be expressed as @xmath24 likewise , the bs serves mss simultaneously in the same frequency band , while each ms feeds back the csi through carrier sense multiple access ( csma ) , which can listen to channel before transmitting a packet to avoid the avoidable collisions .
sender retransmits after some random time if there is a collision . for efficiency , slotted csma
is considered : time is slotted and a packet can only be transmitted at the beginning of one slot .
sender finds out whether transmission was successful or experienced a collision by listening to the ack / nack broadcast from the receiver . without loss of generality
, we consider the slotted @xmath25-persistent csma in @xcite , which can be described by the following steps : * if the channel is idle , transmit with probability @xmath25 , and delay for worst case propagation delay for one packet with probability @xmath26 ; * if the channel is busy , continue to listen until medium becomes idle , then go to step @xmath27 ; * if transmission is delayed by one time slot , continue with step @xmath27 . for slotted @xmath25-persistent csma , the throughput ( @xmath28 ) is given by @xcite : @xmath29)\cdot\exp\big(g(1-p)^{k+1}+\alpha g[-(k+1)+\frac{1-(1-p)^{k+2}}{p}]\big)}{(1+\alpha)\cdot\exp(g(1+\alpha))+\alpha \sum_{k=1}^{\infty}\cdot\exp\big(g(1-p)^k+\alpha g[-k+\frac{1-(1-p)^{k+1}}{p}]\big ) } , \label{eq : ppersist}\end{aligned}\ ] ] where @xmath30 , @xmath31 is the propagation delay , @xmath32 is the packet transmission time , and @xmath33 is the offered load ( overall rate ) . by csma , each user tries to adjust its requested feedback rate @xmath34 over the uplink bandwidth @xmath19 .
however , if the overall rate is too high , due to the random access nature , the network would be congested . as a result
, the accurate rate @xmath35 of user @xmath7 will reduce a lot . from @xcite , we have @xmath36 where the overall rate @xmath37 , and @xmath38 is the maximum network payload .
similarly , the downlink bandwidth can be calculated as that in ( [ eq : bdl ] ) : @xmath39 . in a closed - loop wireless communication system
, the ms needs to feed back the quantized csi back to the bs to perform transmit precoding . for simplicity and without loss of generality
, we here use the equivalent _ quantized feedback channel _ by transforming the real channel matrix in terms of feedback rate and distortion .
we consider a limited and lossless feedback channel . through csi quantization
, the real channel output for the @xmath7-th ms , denoted by @xmath40 , can be modeled as @xcite @xmath41 where @xmath42 represents the quantized feedback channel output with zero mean and @xmath43 variance , @xmath44 is an independent additive noise matrix with each entry corresponding to an i.i.d .
gaussian variable with distribution @xmath45 , and @xmath46 represents the channel quantization distortion constraint .
note that @xmath47 and @xmath48 are mutually independent . due to imperfection in the feedback channel ,
the quality of the feedback information can be measured by the distortion on the source @xmath40 from its representation @xmath49 , which is defined by @xmath50 given distortion rate @xmath46 , the quantized csi can be modeled as @xmath51 where @xmath52 , the elements of @xmath53 are i.i.d .
gaussian variables with distribution @xmath54 , and @xmath55 .
the detailed derivations of @xmath56 and @xmath57 is given in appendix a. based on the shannon s rate - distortion theory of continuous - amplitude sources , the rate - distortion function of a zero - mean and unit variance complex gaussian source is given by @xcite @xmath58 where @xmath34 represents the feedback rate of user @xmath7 .
it can be observed in ( [ eq : rat ] ) that when @xmath59 , i.e. , completely distorted , the feedback rate is equal to zero , while @xmath60 , this requires the infinite feedback rate to realize the undistorted csi . substituting ( [ eq : rat ] ) into ( [ eq : qua ] ) , the quantized csi matrix , @xmath47 , can be expressed as a function of the feedback rate , @xmath34 @xmath61 which clearly connects the feedback rate , @xmath34 , with the quantized csi , @xmath62 , in order to reveal its impact on system performance .
after normalization , ( [ eq : quah ] ) becomes @xmath63 which can be used to perform precoding at the bs side .
in this section , we first present transmit precoder implemented at bs , and then discuss a few properties of limited csi feedback . for simplicity and without loss of generality
, we just consider the conventional mmse based precoder design , and other advanced precoding approaches can be readily applied in this paper . the received signals in ( [ eq : sui ] )
can be rewritten in a matrix form @xmath64 where @xmath65^t\in\mathbb{c}^{n_s{\times}1}$ ] , @xmath66^t\in\mathbb{c}^{n_s{\times}n_t}$ ] , @xmath67\in\mathbb{c}^{n_t{\times}n_t}$ ] , and @xmath68^t\in\mathbb{c}^{n_s{\times}1}$ ] . in this section ,
we demonstrate that the following form for the precoder @xmath69 with two free scalar parameters @xmath70 and @xmath71 , is a general form for an optimal linear precoder , and @xmath72^t\in\mathbb{c}^{n_s{\times}n_t}$ ] . by varying the choice of these parameters ,
optimality can be achieved with respect to a variety of criteria that have been considered in the literature . in general
, @xmath70 is a normalization constant used to comply with the unit transmit power constraint ( averaged over data symbols ) , and it can be expressed as @xmath73 where @xmath74 represents the unnormalized transmit precoder .
the other parameter , @xmath71 , is typically a regularization parameter . maximizing the average sinr under flat fading , which is the same for all mss due to the assumption of symmetry in the distribution of the channel matrix , and taking @xmath5 and @xmath6 ( which in this context , is the number of transmit antennas as well as the number of mss with single antenna ) large
, the optimal @xmath71 can be expressed as the following general form @xcite @xmath75 note that although @xmath71 was solved for the equal sinr case in @xcite , @xmath71 is really a tunable parameter that can be optimized for other criteria . by setting @xmath76 , ( [ eq : uiprege ] ) becomes the simplest and the most common zero - forcing ( channel inversion ) precoder @xmath77 recalling ( [ eq : sui ] ) , the signal to noise plus interference ratio ( sinr ) of the @xmath7-th ms can be written as @xmath78 its corresponding throughput can be expressed as @xmath79 since the precoder @xmath80 is designed by the feedback csi , @xmath81 , optimal performance using the mmse rule can be achieved when @xmath82 , for @xmath83 $ ] .
we in this subsection present a few properties of ( [ eq : snrk ] ) based on the csi feedback : 1 .
@xmath84 is continuous and monotonic increasing in @xmath34 . + it is obvious in ( [ eq : snrk ] ) that @xmath84 is continuous in @xmath34 .
in addition , from ( [ eq : quah ] ) , we can see that the increase of @xmath34 , i.e. the decrease of @xmath46 , improves the amount of the feedback csi , @xmath47 .
this enhances the accuracy of the constructed precoder , @xmath85 , and thus , increases the value of @xmath86 .
@xmath87 , where @xmath88 is a constant .
+ given @xmath89 ( for @xmath83 $ ] , but @xmath90 ) , we have @xmath91 .
hence , @xmath84 converges to a constant when the feedback rate is large enough . in 1 )
, it implies that the feedback rate ( @xmath34 ) of every ms should be as large as possible .
however , if using orthogonal channels in subsection - ii - c-1 ) , this will reduce the downlink bandwidth and then decrease the system throughput . by adopting the csma protocol in subsection - ii - c-2 ) , this will results in more collisions in the uplink , and degrade the effective feedback rate . obviously , there exists a trade - off between feedback rate ( @xmath34 ) and system throughput ( @xmath92 ) in both uplink multiple access protocols .
notice that these properties will be reused to prove the existence of the nfcp equilibrium in subsection - iv - b .
in this section , we first define the utility function
. then , we describe the noncooperative csi feedback control game .
next , we use the pricing method to improve the performance of the proposed game . the convergence to the nash equilibrium is also proved .
finally , we construct a centralized solution for performance comparison .
the concept of utility is commonly used in microeconomics and refers to the level of satisfaction the decision - taker receives as a result of its actions .
mss access a wireless system through the air interface that is a common resource and they transmit information expending bandwidth resources .
since the air interface is a shared medium , each ms transmission is a source of competition for others .
sinr is an effective measure of the quality of signal reception for the wireless user @xcite .
an ms tries to achieve a high quality of reception ( throughput in this paper ) while at the same time expending a certain amount of system bandwidth to feedback csi .
obviously , the csi feedback rate determines the accuracy of the precoder , and thus , affect the system throughput .
therefore , it is possible to view both throughput and the csi feedback rate as commodities that a wireless user desires . the utility function of the @xmath7-th ms can be expressed as @xmath93 utility as defined above is the throughput conditioned on the feedback bit .
note that if @xmath94 , omnidirectional transmission instead of precoding should be used which results in minimum value of @xmath95 .
this suggests that , in order to maximize utility , all users in the system should feed back a certain amount of csi . for orthogonal feedback channels in ( [ eq : bdl ] ) , when the amount of feedback increases , downlink bandwidth , @xmath19 , will decrease , and thus reduce the system throughput .
with regard to csma , when the network payload increases , more collisions happen and consequently the average delay for each packet increases .
any network payload larger than @xmath38 in ( [ eq : csmarate ] ) will cause an unacceptable average delay . as a result ,
the utility becomes zero .
all these facts indicate that the feedback rate should not be either too small or too large for better utility . in other word
, there is an optimal point on how much to feedback from each ms point of view .
intuitively , there exists a tradeoff relationship between obtaining high throughput and requiring small amount of csi feedback on the condition of a total system bandwidth constraint . finding a good balance between the two conflicting objectives is the primary focus of the csi feedback rate control component of radio resource management .
this tradeoff is illustrated through the conceptual plot in fig .
[ fig : utility_tradeoff ] , where orthogonal feedback channel is assumed . if the feedback rate were fixed , the terminal would experience higher throughput as the sinr increases which leads to increased satisfaction of the use of the system resources .
if the sinr were to be fixed ( fixed throughput ) , increasing the feedback rate expedites uplink bandwidth , which effectively reduces the satisfaction of the mobile terminal . for sufficiently large sinr values ,
the throughput approaches zero , which results in an asymptotic decrease in utility in the high sinr region . in the sequel
, we first consider the noncooperative feedback rate control game ( nfc ) where each ms tries to maximize its individual utility .
let @xmath96 $ ] denote the nfc where @xmath97 is the index set for the mobile users currently in the cell , @xmath98 is the strategy set , and @xmath99 is the payoff function of user @xmath7 .
each user selects a feedback rate level @xmath34 such that @xmath100 .
let the feedback rate vector @xmath101 denote the user s strategies in terms of the selected rate levels of all the users , where @xmath102 is the set of all rate vectors .
the resulting utility level for the @xmath7-th user is @xmath103 , where @xmath104 denotes the vector consisting of the other user s strategies other than the @xmath7-th user .
this notation emphasizes that the @xmath7-th user has control over its own rate , @xmath34 only .
the utility of the @xmath7-th ms with feedback rate @xmath34 can be expressed more rigorously as @xmath105 note that ( [ eq : ukr ] ) demonstrates the strategic interdependence between mss .
the level of utility each ms gets depends on its own feedback rate and also on the choice of other players strategies , through the sinr @xmath84 of that user .
the efficiency function can be chosen to represent any precoding scheme described in section iii . in this paper , we assume that the strategy space , @xmath98 , of each user is a compact , convex set with minimum and maximum rate constraints denoted by @xmath106 and @xmath107 , respectively . for simplicity , we let @xmath108 for all @xmath7 , which results in the strategy space @xmath109 $ ] .
the utility function takes the generic form given in fig . 3 for fixed interference plus noise .
the nfc game can be expressed as @xmath110 from ( [ eq : nfc ] ) , it clearly indicates that the feedback rate that optimizes individual utility depends on rates of all the other mss in the network .
it is necessary to characterize a set of rates where the users are satisfied with the utility they receive given the rate selections of other users .
nfc offers a solution to the rate control problem where no ms can increase its utility any further through individual effort .
thus , it is an outcome obtained as a result of distributed decision taking , which could be expected to be less efficient than a possible rate selection obtained through cooperation between terminals and/or as a result of centralized optimization .
specifically , in the nfc , each ms aims to maximize its own utility by adjusting its own feedback rate , but it ignores the cost ( or harm ) it imposes on the other terminals . for example , by orthogonal feedback channels , if one ms increases the usage of bandwidth in order to send more csi , it will decrease the available bandwidth for other mss . while for csma , the higher feedback rate will cause the heavier collision possibility . to overcome this problem , we resort to a usage - based pricing schemes . by introducing a pricing factor for the feedback csi
, we can increase system performance by implicitly inducing cooperation , and yet we maintain the noncooperative nature of the resulting feedback rate of the csi control solution . within the context of a resource allocation problem for a closed - loop wireless system
, the resource being shared is the radio environment , and the resource usage is determined by ms s feedback rate .
hence , efficiency in feedback rate control can be promoted by the proposed usage - based pricing strategy where each user pays a penalty proportional to its usage amount , i.e. rate , of feedback csi .
the nfcp can be expressed as the following optimization problem @xmath111 where @xmath112 $ ] represents a @xmath6 player noncooperative feedback rate control game with pricing ( nfcp ) , @xmath113 is the utility for nfcp , and @xmath114 denotes the pricing function for the @xmath7-th ms , which in this paper is restricted to linear schemes of the form @xmath115 here @xmath116 is announced by the bs and is a constant referred to as the price factor per feedback bandwidth .
note that ( [ eq : mo ] ) also demonstrates the strategic interdependence between users , and the pricing factor needs to be tuned such that user self - interest leads to the best possible improvement in overall network performance . combining ( [ eq : uk ] ) and ( [ eq : pricef ] ) , the nfcp with linear price in ( [ eq : mo ] ) is as follows @xmath117 p100 mm * algorithm 1 : non - cooperative csi feedback rate control game with a given price @xmath116 at the ms side * + 1 .
set initial csi vector at time @xmath118 : @xmath119 . also , let @xmath120 ; \2 .
for all @xmath121 , such that @xmath122 : @xmath123 given @xmath124 , compute : + @xmath125 .
+ by considering the nfcp algorithm in ( [ eq : mof ] ) , given @xmath116 , a sequence of rates can be generated in table [ tab : nfcp_terminal ] .
we refer to @xmath126 as the _ set _ of best feedback rates for the @xmath7-th ms at time @xmath121 instance in response to the interference vector @xmath124 . for the network level algorithm for each value of @xmath116
, we may first run the nfcp when @xmath127 , which is equivalent to the nfc described in ( [ eq : nfc ] ) .
once the equilibrium with no price is obtained , the nfcp is played again after incrementing the price factor , @xmath116 , by a positive value , @xmath128 .
algorithm [ tab : nfcp_terminal ] returns a set of csi rates at equilibrium with this value of the price factor . if the utilities at this new equilibrium with some positive price improve with respect to the previous instance , the price factor is incremented and the procedure is repeated .
we continue until an increase in results in utility levels worse than the previous equilibrium values for at least one user .
we declare the last value to be the best price factor , @xmath129 .
the way that @xmath129 is determined by the network is summarized in algorithmic format in table [ tab : nfcp_network ] .
p100 mm * algorithm 2 : non - cooperative csi feedback rate control game with price algorithm for the whole network * + 1 .
set @xmath127 and announce @xmath127 to all mss ; \2 .
get @xmath130 for all at equilibrium using @xmath131 , increase @xmath132 , and then announce to all mss ; \2 .
if @xmath133 for all then go to step @xmath134 , else stop and declare @xmath135 .
+ in this subsection , we investigate the equilibrium of the proposed games , at which no player can improve its utility by changing its own strategy only .
note that since nfc can be treated as a special case of nfcp when @xmath127 , it would be sufficient to declare the equilibrium existence of the nfc if a nash equilibrium exists in the nfcp .
a rate vector @xmath136 is a nash equilibrium of the npcp @xmath137 $ ] if , for every @xmath138 , @xmath139 , @xmath140 .
there are some existing theorems to show the existence .
we only need to prove that the proposed game satisfies the requirements of the theorems .
it has been shown a nash equilibrium exists , for @xmath141 : a nash equilibrium exists in the nfcp , @xmath142 $ ] if @xmath143 : 1 .
@xmath144 , the support domain of @xmath145 , is a nonempty , convex , and compact subset of a certain euclidean space @xmath146 .
@xmath145 is continuous in @xmath147 and quasi - concave in @xmath144 . obviously , the support domain @xmath144 , which is a vector , satisfies the first condition .
to prove that @xmath148 is quasi - concave , it is equivalent to prove that the first - order derivative of @xmath148 is a monotonic decreasing function whose value varies from positive to negative in term of @xmath144 @xcite . for convenience ,
let @xmath149 and @xmath150 , and thus @xmath151 .
the first - order derivative of @xmath148 with respect to @xmath34 can be calculated as @xmath152 where @xmath153 . as @xmath84 is continuous and monotonic increasing in @xmath34 ,
proved in subsection - iii - c-1 , it is obvious that @xmath154 is also continuous and monotonic increasing in @xmath34 , i.e. @xmath155 .
in addition , since @xmath154 is concave in term of @xmath34 , @xmath156 , which indicates that @xmath157 is a decreasing function . because @xmath158 is monotonic decreasing in @xmath34 , we may conclude that both @xmath159 and @xmath160 are decreasing functions .
hence , @xmath161 is monotonic decreasing in @xmath34 . as @xmath87
, we may easily obtain @xmath162 . from ( @xmath163 ) , given the pricing factor @xmath116 , we can then have : * @xmath164 ; * @xmath165 . hence , @xmath148 is a concave function , and as every concave function is quasiconcave , _ theorem 1 _ is proved .
finally , we prove the existence of the equilibrium of the game , and can conclude that game @xmath166 of ( [ eq : mof ] ) always admits at least one nash equilibrium . to compare the performance ,
a centralized scheme is constructed assuming all csi is known .
the objective is to optimize the sum rate capacity defined in ( [ eq : uk ] ) subject to the constraint of feedback rates : @xmath167 notice that our proposed noncooperative game theoretic based algorithm is distributive , in the sense that only the price information needs to be exchanged , while the centralized scheme needs to gather all the information , which will cause significant signalling .
there are several implementation issues for the proposed scheme .
firstly , the channel estimation for the downlink channel might not be accurate due to both fast fading and noise effects . under this condition
, the transmit precoder formula should be rewritten considering the estimation inaccuracy .
secondly , the proposed scheme needs iteratively update the price and rate information .
a natural question arises if the distributed scheme has less signalling than the centralized scheme .
the comparison is similar to distributed and centralized power control in the literature @xcite .
since the channel condition is continuously changing , the distributed solution only needs to update the difference of the parameters such as rate and pricing factor , while the centralized scheme requires all channel information in each time period . as a result
, the distributed solution has a clear advantage and dominates the current and future wireless network design .
for example ,
the power control for cellular networks , the open loop power control is done only once during the link initialization , while the close - loop power control ( distributed power allocation such as @xcite ) is performed @xmath168 times for umts and @xmath169 times for cdma2000 . finally ,
for the multi - bs multi - ms case , we can use clustering method to divide the network into sub - networks , and then employ the single bs - mss solution proposed in this paper .
in this section , we provide simulation results for the proposed distributed games .
all simulations are performed for a bpsk modulation over the rayleigh fading channels with the mmse precoder in ( [ eq : uiprege ] ) . for simplicity
, we assume that both the transmit power and the noise variance are normalized to unit .
the specific parameters are given below each figure .
in this subsection , we first provide simulations to evaluate the impact of csi feedback on each individual . here
we plot the utility of ms@xmath170 in term of its csi feedback rate by fixing the feedback rates of the rest mss . for simplicity , we assume @xmath134 mss and the csi feedback rate of ms@xmath171 is @xmath172 , and @xmath173 , respectively . from fig .
[ fig : nfcp_ue ] , we can see that when the feedback rates of other users are fixed ( fixed uplink bandwidth ) , the target ms will first experience high throughput as its csi feedback rate increases , which then leads to increased satisfaction of the use of the system resources . for sufficiently large @xmath174 values
, the utility of ms@xmath170 begins to decrease .
it is obvious that the utility function of each ms is a concave function in terms of the feedback rate , which again partially proves through simulations the existence of equilibrium of the proposed nfcp game .
[ fig : nfcp_ms ] is constructed by letting the algorithm in table [ tab : nfcp_terminal ] reach the nash equilibrium at each value of @xmath116 .
the best price factor can be found if all mobile users receive worse overall payoff than the previous equilibrium utility according to algorithm [ tab : nfcp_network ] .
it can be also observed from the figure when the pricing factor increases , the total utility and the sum rates first increase , as shown in the small window , and then begin to decrease .
it indicates that solution by nfcp with @xmath175 offers a significant improvement in total utilities with respect to the nfc when @xmath176 , where pricing factor is not involved . at high pricing factor
, we can see both sum utility and rate converge to a constant value .
this is because the system stops requiring users feedback csi as it costs too much . in fig .
[ fig : nfcp_2ms ] , we compare the proposed nfcp game theoretical approach with the centralized scheme . from the simulation results , we can see that the distributed solution and the centralized solution are asymptotically the same if @xmath116 is in the right region . when @xmath116 is too large , the mss will be reluctant to feedback . when @xmath116 is too small , the mss will feed back in a non - cooperative manner . in fig .
[ fig : nfcp_dl_2ms ] , it shows that the variations of the sum feedback rate as well as the individual feedback rate in term of the pricing factor , where @xmath177 . from the figure
we can see that these results match very well the sum rate result in fig .
[ fig : nfcp_ms ] . when @xmath127 , it requires the maximum amount of feedback .
but with the price increase , the feedback rate starts to decrease until zero , which make the throughput dropped to minimum . for simplicity
, we here consider a special case of slotted @xmath25-persistent csma by setting @xmath178 , i.e. , @xmath27-persistent csma . in fig .
[ fig : csma ] , it shows the total throughput again the traffic load .
it indicates that there exists an optimal transmission rate corresponding to the maximum throughput . in fig .
[ fig : nfcp_csma_ue ] , we examine the impact of csi feedback on each individual using csma . from fig .
[ fig : nfcp_csma_ue ] , we can see that the utility function of each ms is a concave function in terms of the feedback rate , which again proves the effectiveness of the proposed nfcp game . in fig .
[ fig : nfcc_csma_2ms ] , we evaluate the throughput performance in term of the pricing factor .
it shows in fig .
[ fig : nfcc_csma_2ms ] that the proposed nfcp provides much better results than the nfc game .
[ fig : nfcc_csma_c2ms ] compares the proposed nfcp game with the centralized scheme . from the simulation results
, we can see that the distributed solution and the centralized solution are almost the same when @xmath116 is adjusted to the optimal working point .
in this paper , we have studied the csi feedback rate control problem in a single - cell wireless data network , where a multiple - antenna bs communicates with a number of co - channel users through a mmse precoder . specifically , we proposed a non - cooperative feedback - rate control game without and with price .
the price function is a linear function of the csi feedback rate .
the existence of the nash equilibrium of such a game is proved .
simulation results are performed over fdma and csma protocols in the feedback channel .
it shows that the distributed nfcp game with the proposed utility results in improving the overall throughput of wireless data networks , and the simple distributed algorithm can provide comparative performance in comparison of the centralized one by properly varying the pricing parameter .
thought channel quantization , @xmath56 can be simply expressed by the following linear function @xmath179 the real channel output @xmath180 and its corresponding quantized channel @xmath47 in ( [ eq : qua ] ) satisfies the following linear extreme conditions : combining ( [ eq : muvalue ] ) and ( [ eq : nuvalue ] ) , we may easily get @xmath52 . recalling ( [ eq : hk ] ) , from ( [ eq : qua ] ) , we can also have @xmath185=\mathrm{var}\left[\mu\mathbf{h}_{k}+\nu\mathbf{n}_{q}\right ] = \mathrm{var}\left[\mu\mathbf{h}_{k}\right ] + \mathrm{var}\left[\nu\mathbf{n}_{q}\right]\rightarrow 1-d_k=\mu^2+\nu^2 , \label{eq : varrea}\ ] ] and thus , we can finally obtain @xmath55 .
v. tarokh , n. seshadri , and a. r. calderbank , `` space - time codes for high data rate wireless communications : performance criterion and code construction , '' _ ieee trans .
inf . theory _ ,
2 , pp . 744765 , mar .
d. j. love , r. w. heath , v. k. n. lau , d. gesbert , b. d. rao , and m. andrews , `` an overview of limited feedback in wireless communication systems , '' _ ieee j. sel . areas .
13411365 , oct . 2008 .
s. zhou and g. b. giannakis , `` optimal transmitter eigen - beamforming and space - time block coding based on channel mean feedback , '' _ ieee trans .
signal process .
25992613 , oct .
2002 .
m. maskery , v. krishnamurthy , and q. zhao , `` decentralized dynamic spectrum access for cognitive radios : cooperative design of a non - cooperative game , '' _ ieee trans .
2 , pp . 459469 , feb . ,
2009 . c. b. peel , b. m. hochwald , and a. l. swindlehurst , a vector - perturbation technique for near - capacity multi - antenna multi - user communication - part i : channel inversion and regularization , " _ ieee trans .
commun _ , vol .
195 - 202 , jan . 2005
|
it has been well recognized that channel state information ( csi ) feedback is of great importance for dowlink transmissions of closed - loop wireless networks . however , the existing work typically researched the csi feedback problem for each individual mobile station ( ms ) , and thus , can not efficiently model the interactions among self - interested mobile users in the network level . to this end
, in this paper , we propose an alternative approach to investigate the csi feedback rate control problem in the analytical setting of a game theoretic framework , in which a multiple - antenna base station ( bs ) communicates with a number of co - channel mss through linear precoder .
specifically , we first present a non - cooperative feedback - rate control game ( nfc ) , in which each ms selects the feedback rate to maximize its performance in a distributed way . to improve efficiency from a social optimum point of view ,
we then introduce pricing , called the non - cooperative feedback - rate control game with price ( nfcp ) .
the game utility is defined as the performance gain by csi feedback minus the price as a linear function of the csi feedback rate .
the existence of the nash equilibrium of such games is investigated , and two types of feedback protocols ( fdma and csma ) are studied .
simulation results show that by adjusting the pricing factor , the distributed nfcp game results in close optimal performance compared with that of the centralized scheme .
|
a neutron star ( ns ) is born as a result of the core collapse supernova explosion if the initial mass of the main sequence star is around 8 to 20 solar mass ( @xmath3 ) .
various eos , relations between the density and the pressure , are used to describe the interior of neutron stars .
the resulting central density of a neutron star is expected to reach up to several times of the nuclear saturation density ( @xmath4 @xmath5 ) .
hence , a neutron star is one of the best astrophysical laboratories to study the physics of the extremely dense nuclear matter .
figure [ fig : ns_section ] shows the structure of neutron stars in the theoretical point of view . outside the envelope , there exists a very thin atmosphere which is composed of hydrogen and heavy elements @xcite . depending on the temperature , its thickness ranges from a few centimeters to a few millimeters and the thermal radiation is expected to occur in this region .
the envelope plays a role as a thermal insulator @xcite bewteen the outer crust and the atmosphere . in the outer crust ,
heavy nuclei exist with free gas of electrons , and the bcc ( body centered cubic ) structure of nuclei is expected . as the density increases , neutrons drip out of heavy nuclei and can exist as free gas in the inner crust .
the neutron @xmath6 superfluidity can also exist in the inner crust . in the boundary of the outer core and the inner crust
, nuclear pasta phase may appear due to the nature of coulomb interaction @xcite , which favours a special geometrical shape to reduce the free energy . above half of the nuclear saturation density uniform nuclear matter exists .
outer and inner cores occupy roughly 70 % of neutron star s volume . in the outer core
the uniform nuclear matter can exist and neutron @xmath7 and proton @xmath8 superfluidities are expected . in the inner core exotic states with new degrees of freedom , such as kaons , hyperons , and deconfined quarks might appear .
however , the inner structure of neutron star is still unknown due to the theoretical uncertainties , especially at the densities far beyond the nucleon saturation density .
the mass distributions of observed neutron stars are summarized in table 1 of lattimer @xcite .
note that the mass distribution of ns - ns binaries has very sharp peaks since the error bars of mass measurements are relatively small , and the well measured neutron stars in ns - ns binaries have masses @xmath9 . on the other hand , @xmath10
neutron stars have been observed in ns - wd ( white dwarf ) binaries .
this implies that the mass distribution may depend on the binary evolution in addition to the neutron star equation of states @xcite .
recent observations of @xmath11 neutron stars ruled out many soft equations of states with which the maximum mass of neutron star becomes less than @xmath12 @xcite .
another constraint on the nuclear equation of state comes from the x - ray binary observations and their analysis @xcite .
mass distribution in x - ray binaries is very broad without sharp peak indicating that the uncertainties in the mass estimations are very large . in table
[ tab : nsobs ] and fig .
[ fig : obsns ] , we plot the surface temperatures ( @xmath13 ) and photon luminosities ( @xmath14 ) of 18 observed neutron stars .
note that three different models are used to estimate the effective surface temperatures of neutron stars at infinity .
for black body ( bb ) model , the photon luminosity @xmath15 and the effective surface temperature @xmath16 of a neutron star at infinity are related by @xmath17 where @xmath18 is the radius seen by an observer at infinity and @xmath19 is the stefan - boltzmann constant . we can determine @xmath18 by inserting @xmath13 and @xmath14 in the table [ tab : nsobs ] into eq .
( [ eq : lt ] ) . in recent works
@xcite , we have investigated the behavior of nuclear eos at high densities by calculating the mass and radius of neutron stars with several skyrme force models .
we could confirm that the models that satisfy the large mass observation @xcite are consistent with the mass - radius zone in @xcite .
moreover , the conclusion does nt change even if we include exotic degrees of freedom such as kaon condensation @xcite or hyperons @xcite . on the other hand ,
nuclear eos is one of the key ingredients that determine the thermal evolution of neutron stars .
available nuclear models predict very diverse mass - radius relations , so it is well expected that the cooling behavior will be sensitive to nuclear models .
therefore , cooling of neutron stars provides a multi - test ground for the nuclear models ; whether a model good for mass - radius relation is consistent with the observed temperature data or not . in this work ,
we simulate the cooling curve of neutron stars with various eos obtained from both relativistic and non - relativistic models . without the superfluidity effect , the direct urca process is a good indicator whether a specific nuclear model is suitable for neutron star cooling simulation .
if the temperature drop is too fast even for the low mass neutron stars ( @xmath20 @xmath3 ) , such eos is not consistent with most of the neutron star temperature observations . even though the existence of superfluidity highly suppresses neutrino emissivity , once the direct urca process is turned on before the critical temperature
, the superfluidity can not slow down the temperature drop so that the temperature of neutron stars is much below any observational data.thus , the study of cooling curve of neutron star can provide hints about the inner structure of neutron stars and the eos of dense nuclear matter .
in addition to the nuclear eos , physical conditions such as the composition of elements in the envelope , and existence of superfluidity in the core play crucial roles in determining the cooling curve of neutorn stars . in general
, chemical composition of the envelope seldom affects the cooling mechanism , but the surface temperature depends sensitively on whether the elements in the envelope are light or heavy . on the other hand ,
superfluidity directly determines the cooling rate .
if nucleons form a cooper pair and transit to a superfluidic state , the rate of neutrino emissivity is suppressed exponentially .
this may lead to a very slow cooling rate .
however , below a critical temperature , creation and destruction of cooper pairs ignite a fast cooling mechanism , and this can give an abrupt decrease of temperature .
recent literature succeeds to reproduce the cooling curve of cas a in terms of pbf @xcite . in this work , we incorporate pbf to various nuclear models and explore the extent to which the models can reconcile with pbf . the structure of the paper is as follow . in section [ sec : eos ] we summarize the equation of state of nuclear matter which we used in the study of thermal evolution of neutron stars . in this work
, we do not consider exotic matter or quark matter in the core of neutron stars but assume that the core is composed of nucleons ( protons and neutrons ) and leptons ( electrons and muons ) in the form of uniform matter . in section [ sec : cool ] we present the basic equations for neutron star cooling .
simple critical temperature formula is proposed and we analyze the pairing effects to neutron star cooling . in section [ sec : stancool ] , we present the cooling curves for the standard cooling in which superfluidity is neglected with various nuclear models . in section [ sec : super ] , we discuss the effect of the superfluidity to the cooling process . in section [ sec : con ] , we give conclusions from neutron star cooling curves combined with varios eos . in appendix , we summarize numerical methods to solve the diffusion equations for the thermal evolution of neutron stars . .thermal emission from isolated neutron stars .
temperatures were obtained using three different models ; hydrogen atmosphere ( ha ) , magnetized hydrogen atmosphere ( mha ) , and black - body ( bb ) model . for sources with no .
2 @xmath21 7 , two different models were used to link the effective temperature and the photon luminosity .
sources with no .
16 @xmath21 17 have the limited observational data , thus have only one limit . + [ cols="^,<,^,^,^,^,^,^",options="header " , ] in fig .
[ fig : time_step ] , we compare results from three different numerical methods . if the superfluidity happens , penta - diagonal method needs a smaller size of time step to make the result similar with the ones from both the tri - diagonal methods ( even @xmath22 and odd @xmath23 ) and the tri - diagonal methods in which @xmath22 and @xmath23 are defined in all grid points .
b. demorest , t. pennucci , s. m. ransom , m. s. e. roberts , and j. w. t. hessels , nature * 467 * ( 2010 ) 1081 . j. antoniadis _
et al_. , science * 340 * ( 2013 ) 448 .
a. w. steiner , j. m. lattimer , and e. f. brown , apj * 722 * ( 2010 ) 33 .
k. oyamatsu , nucl .
a * 561 * ( 1993 ) 431 . j. m. lattimer , ann .
nucl . part .
* 62 * ( 2012 ) 485 . c. -h .
lee and h. -s .
cho , nucl .
phys . a * 928 * ( 2014 ) 296 .
y. lim , k. kwak , c. h. hyun , and c .- h .
lee , phys .
c * 89 * ( 2014 ) 055804 . g. g. pavlov and v. e. zavlin , _ thermal radiation from cooling neutron stars _ , in texas tuscany .
xxi texas symposium on relativistic astrophysics , edited by r. bandiera , r. maolino , and f. mannucci , singapore , world scientific , 2003 , pp .
319 - 328 [ arxiv : astro - ph/0305435 ] .
m. bender , p .- h .
heenen , and p .- g .
reinhard , rev .
* 75 * ( 2003 ) 121 .
a. w. steiner , m. prakash , j. m. lattimer , and p. j. ellis , phys .
* 411 * ( 2005 ) 325 .
h. mller and b. serot , arxiv : nucl - th/9603037 .
e. chabanat , p. bonche , p. haensel , j. meyer , and f. schaeffer , nucl .
phys . a * 635 * ( 1998 ) 231 .
j. m. lattimer , c. j. pethick , m. prakash , and p. haensel , phys .
* 66 * ( 1991 ) 2701 .
b. l. friman and o. v. maxwell , apj * 232 * ( 1979 ) 541 .
d. g. yakovlev and k. p. levenfish , a&a * 297 * ( 1995 ) 717 .
a. d. kaminker , c. j. pethick , a. y. potekhin , v. thorsson , and d. g. yakovlev , a&a * 343 * ( 1999 ) 1009 .
d. g. yakovlev , k. p. levenfish , and yu . a. shibanov , phys .
* 42 * ( 1999 ) 737 .
w. l. slatter , g. d. doolen , and h. e. dewitt , phys .
a * 26 * ( 1982 ) 2255 .
h. dewitt , w. slattery , and g. chabrier , physica b * 228 * ( 1996 ) 21 .
k. a. van riper , apj * 75 * ( 1991 ) 449 .
d. page , phd thesis , _ neutron star cooling : kaon condensation and nucleon superfluidity_. d. g. yakovlev and c. j. pethick , ann .
. astrophys .
* 42 * ( 2004 ) 169 .
j. m. ziman , _ electrons and phonons _ , oxford university press , oxford , 1960 .
d. a. baiko , p. haensel , and d. g. yakovlev , a&a * 374 * ( 2001 ) 151 .
o. y. gnedin and d. g. yakovlev , nucl .
phys . a * 582 * ( 1995 ) 697 .
s. shternin and d. g. yakovlev , phys .
d * 74 * ( 2006 ) 043004 .
b. posselt , s. b. popov , f. haberl , j. trmper , r. turolla , r. neuhuser , and p. a. boldin , a&a * 482 * ( 2008 ) 617 .
e. h. gudmundsson , c. j. pethick , and r. i. epstein , apj * 272 * ( 1983 ) 286 .
j. m. lattimer and m. prakash , phys . rep .
* 442 * ( 2007 ) 109 .
j. m. lattimer and y. lim , apj * 771 * ( 2013 ) 51 .
a. y. potekhin , g. chabrier , and d. g. yakovlev , a&a * 323 * ( 1997 ) 415 .
d. page and j. h. applegate , apj * 394 * ( 1992 ) l17 .
d. g. yakovlev , a. d. kaminker , and k. p. levenfish , a&a * 343 * ( 1999 ) 650 .
d. page , j. m. lattimer , m. prakash , and a. w. steiner , apj * 707 * ( 2009 ) 1131 .
a. akmal , v. r. pandharipande , and d. g. ravenhall , phys .
c * 58 * ( 1998 ) 1804 .
w. zuo , z. h. li , g. c. lu , j. q. li , w. scheid , u. lombardo , h .- j .
schulze , and c. w. shen , phys .
b * 595 * ( 2004 ) 44 .
a. schwenk , b. friman , and g. e. brown , nucl .
phys . a * 713 * ( 2003 ) 191 .
elgaroey , l. engvik , m. hjorth - jensen , and e. osnes , phys .
* 77 * ( 1996 ) 1428 .
a. d. kminker , p. haensel , and d. g. yakovlev , a&a * 373 * ( 2001 ) l17 .
w. zuo , c. x. cui , u. lombardo , and h. -j .
schulze , phys .
c * 78 * ( 2008 ) 015805 .
l. amundsen and e. stgaard , nucl .
phys . a * 442 * ( 1985 ) 163 .
k. g. elshamouty , c. o. heinke , g. r. sivakoff , w. c. g. ho , p. s. shternin , d. g. yakovlev , d. j. patnaude , and l. david , apj * 777 * ( 2013 ) 22 .
e. h. gudmundsson , c. j. pethick , and r. i. epstein , apj * 259 * ( 1982 ) l19 .
|
we investigate the effects of the nuclear equation of state ( eos ) to the neutron star cooling .
new era for nuclear eos has begun after the discovery of @xmath0 neutron stars psr j1614@xmath12230 and psr j0348@xmath20432 @xcite .
also recent works on the mass and radius of neutron stars from low - mass x - ray binaries @xcite strongly constrain the eos of nuclear matter . on the other hand , observations of the neutron star in cassiopeia a ( cas a ) more than 10 years confirmed the existence of nuclear superfluidity @xcite .
nuclear superfluidity reduces the heat capacities as well as neutrino emissivities . with nuclear superfluidity the neutrino emission processes
are highly suppressed , and the existence of superfluidity makes the cooling path quite different from that of the standard cooling process .
superfluidity also allows new neutrino emission process , which is called ` pair breaking and formation'(pbf ) .
pbf is a fast cooling process and can explain the fast cooling rate of neutron star in cas a. therefore , it is essential to add the superfluidity effect in the neutron star cooling process . in this work ,
we simulate neutron star cooling curves using both non - relativistic and relativistic nuclear models .
the existence of too early direct urca process shows that some of nuclear models do not fit for the cooling simulation .
after this first selection process , the nuclear pairing gaps are searched using the observational neutron star s age and temperature data .
neutron star , nuclear equation of state , neutron star cooling , nuclear superfluidity # 1 # 1
|
since the observation of large transverse polarization of produced @xmath0 hyperons in the inclusive reactions @xmath2 @xcite and @xmath3 @xcite in the middle of the 1970 s , there have been many experimental and theoretical investigations aimed at understanding this striking polarization phenomenon @xcite .
the polarization measurements of @xmath0 hyperons produced in these inclusive reactions have been performed in fixed target experiments , and the data showed that the @xmath0 polarizations are large only for large @xmath4 .
this poses a problem if one wants to investigate this observable further using high energy colliders such as rhic , tevatron or lhc , where the capabilities to measure @xmath0 polarization are restricted to the midrapidity region , where @xmath4 is very small .
however , in this paper we point out a way to by - pass this problem by making a less inclusive measurement : to select two - jet events and to measure the jet momenta @xmath5 and @xmath6 in addition to the momentum @xmath7 and polarization @xmath8 of the @xmath0 that is part of either of the two jets .
an asymmetry proportional to @xmath9 can then arise , which is neither power suppressed , nor needs to be zero the @xmath0 polarization needs to vanish at @xmath10 due to symmetry reasons . in @xmath11 nuclear effects could allow for a nonvanishing asymmetry at midrapidity .
no such effects have been observed thus far .
] at vanishing @xmath4 of the @xmath0 .
the observable is proportional to @xmath12 in the center of mass frame of the two jets . within a factorized description of the process under consideration ,
such an asymmetry can arise from spin and transverse momentum dependent fragmentation functions , which in turn can arise from two types of interactions .
we will point out how the measurement of the suggested asymmetry in principle can allow for a differentiation between these two types of mechanisms . in this way
also hadron collider experiments can contribute to our understanding of the underlying mechanisms that lead to @xmath0 polarization , in addition to experiments with electron - positron and lepton - hadron collisions .
it is well - known that transverse momentum dependent distribution and fragmentation functions , nowadays commonly referred to as tmds , can have a nontrivial spin dependence and that the so - called `` @xmath13-odd '' tmds can lead to single spin asymmetries @xcite .
they are also often referred to as `` naively @xmath13-odd '' , because the appearance of these functions does not imply a violation of time - reversal invariance .
the sivers distribution function @xmath14 , schematically depicted in fig.[siversdistribution ] , is the oldest example of such functions .
[ cc][cc]@xmath15 [ lc][lc]@xmath16 [ lc][lc]@xmath17 [ cc][cc]@xmath18 [ cc][cc]@xmath19 [ cc][cc]@xmath20 [ cc][cc]@xmath21 + it describes the difference between the momentum distributions of quarks inside protons transversely polarized in opposite directions . the sivers effect was put forward @xcite as a possible explanation for the large single spin asymmetries observed in @xmath22 experiments @xcite .
furthermore , it generates single spin asymmetries in semi - inclusive dis @xcite , which have also been measured to be nonzero @xcite , and it results e.g. in asymmetric di - jet correlations in @xmath23 @xcite , which however are not yet visible in the data analyzed @xcite . the fragmentation analogue of the sivers distribution function is called @xmath24 @xcite .
it describes the distribution of transversely polarized spin-1/2 hadrons , such as @xmath0 s , inside the jet of a fragmenting _
unpolarized _ quark , cf .
[ siversfragmentation ] . for this reason
it has been referred to as `` polarizing fragmentation function '' in ref .
@xcite .
[ cc][cc]@xmath25 [ cc][cc]@xmath16 [ cc][cc]@xmath17 [ rc][rc]@xmath26 [ cc][cc]@xmath21 + it is an odd function of the transverse momentum of the observed hadron w.r.t . the quark direction , or equivalently , the jet direction .
despite the similarity between the definitions of @xmath24 and @xmath14 , there are some important differences .
nonvanishing @xmath13-odd distribution functions require soft gluonic interactions between the target remnants and the active partons @xcite .
these interactions can be resummed into wilson lines ( gauge links ) , ensuring the gauge invariance of the operator definitions of the distribution functions @xcite . on the other hand
, there are two mechanisms to generate @xmath13-odd effects in the fragmentation process : through final - state interactions within the jet , e.g. between the observed outgoing hadron and the rest of the jet @xcite , or through soft gluonic interactions between the jet and the hard scattering part ( sometimes also referred to as final - state interactions ) , see figs [ toddfragmentation]a and [ toddfragmentation]b respectively .
final - state interactions : + ( -39,0)(0,-1)18 + @xmath24 + ( a ) soft gluon interactions ( gauge links ) : + ( -39,0)(0,-1)18 + @xmath27 + ( b ) as for the distribution functions , the latter interactions give rise to wilson lines . both effects can be expressed in terms of a tmd fragmentation function . the @xmath0 polarization observable that is the subject of this paper in principle allows for a differentiation between these two effects .
there is considerable interest in this issue , as it could shed light on the color flow dependence of single spin asymmetries and the ( non)universality of transverse momentum dependent fragmentation functions . in recent years
it has become apparent that @xmath13-odd tmds enter in different ways in different processes , depending on the color flow in the partonic subprocesses - even tmds are not excluded , they will not be considered here . ] .
the sign relation between the sivers function appearing in semi - inclusive dis and in drell - yan scattering was the first example of such process dependence @xcite .
more complicated relations were discussed soon afterwards for hadron production processes in hadronic collisions @xcite .
process dependence may seem at odds with factorization @xcite , but since the color flow dependence can be explicitly taken into account through the determination of the wilson line structure , the process dependence is explicitly calculable and hence , predictive power may be retained albeit in a less straightforward manner . for the specific example of the single spin asymmetry in @xmath23 the effects of color flow dependence
have been taken into account @xcite , leading to a suppression w.r.t . the standard parton model expectation .
for fragmentation functions the situation is complicated further by the fact that only one of the two mechanisms leading to nonzero @xmath13-odd fragmentation functions depends on the color flow structure @xcite .
however , a claim based on a model calculation @xcite and on more general arguments @xcite has been put forward that for @xmath13-odd fragmentation functions no color flow dependence arises at all .
this makes an experimental test of this dependence , or of its absence , all the more interesting .
we propose to measure it in the process @xmath1 , where the brackets indicate that the @xmath0 is part of one of the two observed jets which are almost back - to - back in the plane perpendicular to the beam axis and its polarization is measured .
this measurement can be done for instance at lhc and rhic , although as we will point out the situation at lhc is more straightforward thanks to the dominance of gluon - gluon scattering in the region of interest . the @xmath0 polarization observable we put forward here is similar to the sivers effect observable in @xmath28 @xcite , but the @xmath0 polarization observable does not depend on a non - collinearity of jets in the transverse plane .
in other words , even if one were to assume the initial partons to be collinear to the parent hadrons , in which case the jets are exactly back - to - back in the plane perpendicular to the beam axis , an asymmetry can still arise .
therefore , it is expected that the @xmath0 polarization observable is less sensitive to effects that would smear out the partonic @xmath29 kinematics .
this prompts us to ignore contributions that depend on nontrivial spin and transverse momentum dependent effects in the initial states ( which would be proportional to @xmath30 in the notation of ref .
@xcite ) .
the observable turns out to be very sensitive to the unpolarized @xmath0 fragmentation functions .
the relative importance of contributing partonic subprocesses differs strongly depending on whether one uses for instance the fragmentation functions by de florian et al .
@xcite ( fsv ) or by albino et al .
@xcite ( akk ) . therefore ,
even if no polarization effect is seen , the unpolarized data obtained by the proposed measurement can in any case lead to a considerable improvement in the determination of the unpolarized @xmath0 fragmentation functions .
we will explain why it is necessary to measure both jet momenta and also how it differs from the process @xmath31 .
we will start with a discussion of that process .
anselmino et al .
@xcite have attempted to describe the latter in terms of @xmath24 , but although a satisfactory description of the @xmath0 polarization at larger values of the transverse momentum of the @xmath0 could be obtained , considerable doubts about the applicability of the assumed factorization have arisen afterwards .
a similar problem is unlikely to arise for the @xmath0 polarization study in the process @xmath1 at lhc and rhic .
the @xmath0 s measured in the process @xmath31 @xcite by the fixed target experiments have transverse momenta @xmath32 w.r.t .
the beam direction ranging up to 4 gev@xmath33 .
the @xmath0 polarization transverse to the production plane first grows linearly with @xmath32 , but around @xmath34 gev@xmath33 it levels off and remains approximately constant . for large @xmath32
one expects the polarization to fall off as @xmath35 , based on theoretical ( collinear expansion ) arguments @xcite . the high energy collider experiments would be ideally suited to demonstrate this power - law fall - off behavior , if it were not for the small @xmath4 values probed in those experiments .
the linearly rising behavior at very low @xmath36 is well - described by recombination models employing spin - orbit couplings @xcite .
however , the plateau region for @xmath37 gev@xmath33 is less straightforward to incorporate in such soft physics models .
anselmino et al .
@xcite assumed that the @xmath32 in this region may be considered a hard scale which then would allow for factorization of the cross section expression .
however , since the scale is not too large either , instead of restricting to collinear factorization , transverse momentum dependence at the parton level was included . in this way the @xmath0 polarization could be described in terms of the fragmentation function @xmath24 .
although a satisfactory description of the higher-@xmath32 data was obtained in this way , a preliminary study of the unpolarized cross section within the same factorized approach resulted in considerable disagreement with the data , as mentioned in ref .
@xcite . apparently , at large @xmath4 a @xmath32 value of 3 or 4 gev@xmath33 is not yet large enough to allow for a factorized description in the case of @xmath0 production , as opposed to for instance pion production .
inclusion of intrinsic transverse momentum in the initial state parton distributions may lead to less discrepancy with the data , as studied for pion production @xcite , but remains to be done for @xmath0 production . despite the doubts about whether @xmath24 can be extracted from the existing fixed - target @xmath0 polarization data , the collider experiments at rhic and lhc are in a kinematic region that should allow for a factorized description of the cross section and
thus can be considered safe .
as mentioned , they do have the drawback that @xmath0 measurements are generally restricted to small values of @xmath4 , where the @xmath0 polarization in @xmath31 will be close to zero .
therefore , we next discuss a different process , where this drawback does not appear . the main difference between the traditional process @xmath31 and the one considered in this paper , @xmath38 comes from the way @xmath24 enters in the asymmetry expressions . in the latter process it enters in an unintegrated way ,
i.e. the asymmetry is proportional to @xmath39 , where @xmath40 and @xmath41 are observed . in the traditional process on the other hand
, there is an integration over @xmath41 , which implies that one essentially probes the first derivative of the partonic cross section ( in case of collinear initial partons ) , which results in the aforementioned @xmath42 dependence . in @xmath38
this @xmath42 factor does not arise .
we consider the process @xmath43 .
as indicated by the brackets , the @xmath0 is part of the jet with observed momentum @xmath44 , where the two jets are almost back - to - back in the plane perpendicular to the beam axis .
the projections of the outgoing momenta in this plane will be denoted by @xmath45 , @xmath46 and @xmath47 .
we will also use the scaled variables @xmath48 and @xmath49 . at leading order the hadronic process
is mediated by two - to - two partonic scattering @xmath50 . since in the process under consideration both jet momenta
are observed , the momenta of the outgoing partons are known , i.e. @xmath51 and @xmath52 . for the parton momenta
it is advantageous to make the sudakov decompositions @xmath53 for incoming partons and @xmath54 for the outgoing parton that will fragment into the hyperon .
the @xmath55-vectors are lightlike four - vectors that have nonzero overlaps with the associated hadron momentum ( i.e. @xmath56 and @xmath57 .
there is a certain degree of arbitrariness in the choice of these @xmath55-vectors , but it is required that @xmath58 and @xmath59 are suppressed by inverse powers of one of the hard scales in the process ( such as @xmath60 or @xmath61 ) .
this requirement ensures in the first place that the cross section can be expressed in terms of tmds , i.e. functions of only the momentum fractions and intrinsic transverse momenta . by taking @xmath55 to be a measurable four - vector in the process , e.g. @xmath62 or one of the incoming hadron momenta @xmath63 ( or any linear fixed combination of them provided @xmath64 is large ) ,
one ensures that the observed momentum fraction @xmath40 and intrinsic transverse momentum squared @xmath65 are lorentz invariant .
moreover , for any of these choices the same @xmath40 and @xmath65 result , up to power corrections , which are neglected in our treatment . below we will choose @xmath66 , for which @xmath67 is always small , of order hadronic scale squared divided by the partonic center of mass energy squared .
the momentum fraction and intrinsic transverse momentum of the outgoing parton are then @xmath68 note that @xmath41 refers to the transverse component of the parton s momentum w.r.t . the @xmath0 direction . the transverse component @xmath69 of the @xmath0 w.r.t .
the jet direction is @xmath70 ( note that the subscripts @xmath13 of @xmath69 and @xmath41 refer to different frames @xcite ) .
the tmd fragmentation functions are commonly written with this vector as a variable .
for the cross section we take as a starting point the assumption that for each observed hadron separately the soft physics factorizes from the hard physics : @xmath71 with @xmath72 .
the convolutions @xmath73 indicate the appropriate dirac and color traces for the partonic hard squared amplitude @xmath74 .
the momentum conservation delta function can furthermore be written as @xmath75 where the @xmath76 are the pseudorapidities of the outgoing jets and @xmath77 denote polar angles in the c.o.m .
frame of the two incoming hadrons .
the @xmath78 indicate the projections of the intrinsic transverse momenta @xmath79 of the incoming partons in the plane perpendicular to the beam axis . a factorization theorem as in eq .
( [ crosssec ] ) has not been proven rigorously for this process , but we believe that it is plausible to assume that if it exists , it will be of the schematic form of the expression in eq .
( [ crosssec ] ) , possibly up to soft factors . in principle
there are several effects that contribute to the spin dependence of the cross section in eq .
( [ crosssec ] ) , viz . the boer - mulders effect and the sivers fragmentation effect . in the boer - mulders mechanism @xcite
the single spin asymmetry arises as a consequence of a correlation between the transverse momentum and transverse spin of a quark inside an unpolarized initial hadron .
hence , it relies on a @xmath41-effect in the initial state that will have to translate into an asymmetry in the final - state . through the hard partonic scattering there
can be @xmath41-smearing , leading to a dilution of the asymmetry in the final state .
we will , therefore , assume that the boer - mulders effect is less important than the sivers fragmentation effect .
the latter is a @xmath41-effect that is directly in the final state , such that there is no dilution from @xmath41-smearing in the partonic scattering .
in addition , one may expect that the chiral - odd contributions ( @xmath80 ) are generally smaller than the chiral - even ones ( @xmath81 ) , similar to what was found in ref .
@xcite for the process @xmath82 using positivity bounds .
based on these considerations , we will simply neglect the intrinsic transverse momenta @xmath79 of the incoming particles , such that only the sivers fragmentation mechanism contributes . as mentioned in the introduction , this effect is described by the tmd fragmentation function @xmath83 , which enters the expression in eq .
( [ crosssec ] ) through the parameterization of the fragmentation correlator @xcite ( suppressing wilson lines and with the convention @xmath84 ) [ fragcor ] @xmath85\\ & = \big(\,d_1(z{,}k_{\lambda\,t}^2 ) -d_{1t}^\perp(z{,}k_{\lambda\,t}^2 ) \frac{\epsilon^{\mu\nu\rho\sigma}k_{\lambda\,\mu } k_\nu s_{\lambda\,\rho } n_\sigma } { m_\lambda\,(k_\lambda{\cdot}n)}\,\big)\,\slash k_\lambda + \text{other functions}\label{fragcorb}\displaybreak[0]\\ & = \big(\,d_1(z{,}k_{\lambda\,t}^2 ) + d_{1t}^\perp(z{,}k_{\lambda\,t}^2 ) \frac{(\hat{\boldsymbol k}{\times}\boldsymbol k_{\lambda}){\cdot } \boldsymbol s_\lambda}{z\,m_\lambda}\,\big)\,\slash k_\lambda + \text{other functions}\ , \label{fragcorc}\end{aligned}\ ] ] in the case of quark fragmentation .
here @xmath17 denotes the quark momentum , which equals the jet momentum @xmath44 , and @xmath55 is the sudakov vector discussed before . the last line ( [ fragcorc ] )
holds in any reference frame where @xmath86 and @xmath87 point in opposite directions .
sometimes the notation @xmath88 is used to indicate the sivers fragmentation function @xcite . as mentioned in the introduction
, there is a complication specifically associated with tmd fragmentation functions . the sivers fragmentation function of eqs . and is actually a combination of the two functions in fig
. [ toddfragmentation ] , arising from two different types of interactions . in the hadronic cross section
the function @xmath89 generated by the final - state interactions will appear convoluted with the usual partonic scattering cross sections .
however , due to the dependence of the soft gluon interactions on the partonic subprocesses ( in particular on its color flow structure ) , the tmd fragmentation function @xmath90 generated by this effect will appear convoluted with modified partonic scattering cross sections .
the contribution of the sivers fragmentation mechanism to the hadronic cross section will thus have the form @xmath91d}}{d\hat t}\ , \widetilde d_{1t}^{c\,\perp}(z{,}k_{h\,t}^2)\ , \ ] ] where the @xmath92 are the usual partonic scattering cross sections and the @xmath93d}$ ] are the modified partonic scattering cross sections ( referred to as gluonic pole cross sections - odd gluon correlators in principle two distinct gluonic pole cross sections @xmath94 and @xmath95 appear .
however , the latter cross section vanishes for the gluon - gluon scattering contribution @xcite , the only contribution considered in sec.[phenom ] .
therefore , the function @xmath96 in this paper refers to @xmath97 . ] ) calculated in refs .
therefore , the sivers fragmentation functions in eqs . and should be read as @xmath98}\widetilde d{}_{1t}^\perp$ ] , where the @xmath99}$ ] are color fractions that depend on the particular term in the partonic squared amplitude @xmath100 which is given by the cut feynman diagram @xmath25 .
these factors are perturbatively calculable and , once absorbed in the partonic scattering cross sections , lead to the gluonic pole cross sections .
the gluonic pole cross sections , then , are gauge invariant weighted sums of cut feynman diagrams @xmath25 with the color fractions @xmath99}$ ] as weight factors . after neglecting intrinsic transverse momentum effects for the incoming particles the sivers fragmentation contribution to the cross section eq
. becomes @xmath101d}}{d\hat t}\ , z\widetilde d{}_{1t}^{c\,\perp}(z{,}(k_\lambda{-}zk_j)^2)\,\bigg]\,\bigg\}\ , \label{cs1}\end{gathered}\ ] ] where the momentum fractions @xmath102 and @xmath103 have been fixed by the first two delta functions on the r.h.s . of eq . .
the partonic scattering cross sections @xmath92 and the gluonic pole cross sections @xmath93d}$ ] are functions of the mandelstam variables @xmath104 , @xmath105 and @xmath106 or , alternatively , of @xmath107 and the variable @xmath108 defined through @xmath109 note that the expression of @xmath108 in terms of @xmath110 and @xmath111 in eq
. holds in the center of mass frame of the initial hadrons .
since we are ignoring masses w.r.t .
@xmath61 , pseudorapidity and rapidity are the same and since differences of rapidities are invariant under boosts , eq .
( [ ydef ] ) also holds in the jet - jet c.o.m .- frame .
it should be emphasized that eq .
( [ cs1 ] ) does not hold in the c.o.m .- frame of the incoming hadrons .
( [ cs1 ] ) forms the central result of this paper .
any nonzero measurement of the @xmath112 asymmetry implies a nonzero sivers effect in the fragmentation process and will thus be very interesting .
it will imply nonzero @xmath24 and/or @xmath27 . due to the different partonic cross sections appearing in front of these two different sivers fragmentation functions , measuring the @xmath108 dependence of the asymmetry in principle offers the possibility to separate the two contributions , and to learn about their relative importance .
this we will explore further in the next section .
if one were to study instead the process @xmath113 without constructing @xmath62 , then one does not have the full partonic kinematic information and one will be probing the combination @xmath114d}/d\hat t}{d\hat\sigma_{ab\rightarrow
cd}/d\hat t}\,\ , \widetilde d_{1t}^{c\,\perp}(z{,}k_{h\,t}^2)\ , \ ] ] averaged over @xmath108 .
in addition , referring to a correlation @xmath112 in the jet - jet c.o.m .- frame , involving three - vectors ( corresponding to the choice @xmath115 ) would not make sense .
one could for instance choose @xmath116 and consider this correlation in terms of three - vectors in any frame in which @xmath117 and @xmath118 point in opposite directions
. in the remainder of this article we will focus on the case where @xmath62 is measured and present results in the jet - jet c.o.m .- frame .
one can decompose the hadronic cross section in a spin - independent and a spin - dependent part : @xmath119 comparing to eq .
it follows that within our approximations the ratio of the spin - dependent and spin - independent cross sections is given by @xmath120d}\ , \widetilde d{}_{1t}^{c\,\perp}\big(z{,}(k_\lambda{-}zk_j)^2)\,\big\ } } { \sum\limits_{a,{\cdots},d } x_1f_1^a(x_1)\,x_2f_1^b(x_2)\,d\hat\sigma_{ab\rightarrow cd}\ , d_1^c\big(z{,}(k_\lambda{-}zk_j)^2)}\ . \label{ratio}\end{aligned}\ ] ] this ratio appears in the single spin asymmetry @xmath121 the relevant partonic hard functions for our study are @xcite @xmath122g}}{d\hat t } = -\frac{\pi\alpha_s^2}{\hat s^2}\frac{2n^2}{n^2{-}1 } \frac{(\hat s^2{+}\hat t^2{+}\hat u^2)^2}{4\hat t^2\hat u^2}\ , \\ & \frac{d\hat\sigma_{qg\rightarrow qg}}{d\hat t } = -\frac{\pi\alpha_s^2}{\hat s^2 } \frac{\hat s^2{+}\hat u^2}{2\hat s\hat u } \big(\,\frac{\hat s^2{+}\hat u^2}{\hat t^2}{-}\frac{1}{n^2}\,\big)\ , & & \frac{d\hat\sigma_{qg\rightarrow[q]g}}{d\hat t } = \frac{\pi\alpha_s^2}{\hat s^2 } \frac{\hat s^2{+}\hat u^2}{2\hat s\hat u } \big(\,\frac{\hat s^2}{\hat t^2}{- } \frac{n^2{+}1}{n^2{-}1}\big(\,\frac{\hat u^2}{\hat t^2 } { -}\frac{1}{n^2}\,\big)\,\big)\ .\mspace{-30mu}\end{aligned}\ ] ] all cross sections are given for massless partons .
the cross sections with the initial or final state particles interchanged are obtained from these by a @xmath123 substitution .
we also define the ratios ( displayed in fig .
[ abfuncts]a ) @xmath124 g } } { d\hat\sigma_{gg\rightarrow gg } } = -\frac{\hat s^2}{2}\,\frac{\hat s^2{+}\hat t^2{+}\hat u^2 } { \hat s^4{+}\hat t^4{+}\hat u^4 } = -\frac{1}{2}\ , \frac{1\,{+}\,y^2\,{+}\,(1{-}y)^2}{1\,{+}\,y^4\,{+}\,(1{-}y)^4}\ , \\
a_q(y ) = \frac{d\hat\sigma_{qg\rightarrow[q]g}}{d\hat\sigma_{qg\rightarrow qg } } = \frac{n^2{+}1}{n^2{-}1 } -\frac{2n^2}{n^2{-}1 } \frac{\hat s^2}{\hat s^2{+}\hat u^2{-}\frac{1}{n^2}\hat t^2 } = \frac{n^2{+}1}{n^2{-}1 } -\frac{2n^2}{n^2{-}1 } \frac{1}{1{+}(1{-}y)^2{-}\frac{1}{n^2}y^2}\ , \\
b(y ) = \frac{d\hat\sigma_{qg\rightarrow qg}}{d\hat\sigma_{gg\rightarrow gg } } = -\frac{n^2{-}1}{2n^2 } \frac{\hat s\hat u ( \hat s^2{+}\hat u^2)(\,\hat s^2{+}\hat u^2{-}\frac{1}{n^2}\hat t^2\ , ) } { ( \hat s^4{+}\hat t^4{+}\hat u^4)(\hat s^2{+}\hat t^2{+}\hat u^2 ) } = \frac{n^2{-}1}{2n^2 } \frac{(1{-}y)(1{+}(1{-}y)^2)(\,1{+}(1{-}y)^2{-}\frac{1}{n^2}y^2\ , ) } { ( 1{+}y^4{+}(1{-}y)^4)(1{+}y^2{+}(1{-}y)^2)}\ .\end{gathered}\ ] ] [ cc][cc]@xmath108 ( a ) [ cc][cc]@xmath110 ( b ) in fig . [ comparison ]
we show the expressions @xmath125 appearing in the denominator of eq .
( [ ratio ] ) , integrated over the pseudorapidity ranges @xmath126 , normalized to the corresponding quantity for gluon - gluon scattering ( fig .
[ comparison]a ) and quark - gluon scattering ( fig .
[ comparison]b ) .
the pseudorapidity integration range is inspired by the @xmath127 coverage of the alice detector at lhc , which has the particle identification capabilities that allow to measure @xmath0 s with transverse momenta @xmath32 of several gev@xmath33 .
we have taken @xmath128 tev as is relevant to the lhc situation and for the unpolarized distribution functions we have used the cteq5l parametrizations evaluated at the scale @xmath129 gev .
the figures do not change significantly with the scale in this region ( see figure [ comparison ] ) .
what we infer from the figure is that the combination @xmath125 which determines the inclusive cross section is dominated by the gluon - gluon ( @xmath130 ) scattering process in this particular kinematic region .
the quark - gluon ( @xmath131 ) scattering contribution is on the order of @xmath132 of the ( @xmath130 ) contribution at perpendicular scales @xmath133 typical for lhc , if we consider jets with transverse momenta between @xmath134 and @xmath135 gev@xmath33 .
similarly , the other partonic contributions are of the order of @xmath132 ( or less ) of the @xmath131 contribution ( fig .
[ comparison]b ) .
[ cc][cc]@xmath133 ( a ) ( b ) whether or not the unpolarized @xmath0 production cross section @xmath136 can also be approximated by the gluon - gluon scattering contribution , depends heavily on the @xmath0 fragmentation functions employed .
several such functions have been fitted to experimental data @xcite , unfortunately with vastly different results for the ratios @xmath137 for the various quark and antiquark flavors . if one restricts to the subprocesses @xmath130 and @xmath131 , then the unpolarized cross section @xmath136 contains @xmath138 with @xmath139 } { x_1f_1^g(x_1)\,x_2f_1^g(x_2)\,d\hat\sigma_{gg\rightarrow gg}\,d_1^g(z ) } \nonumber\\ & \mspace{200mu } + \frac{\sum_q\,x_1f_1^g(x_1)\,x_2f_1^q(x_2)\,\left[\ , d\hat\sigma_{gq\rightarrow qg}\,d_1^q(z ) { + } d\hat\sigma_{gq\rightarrow gq}\,d_1^g(z)\,\right ] } { x_1f_1^g(x_1)\,x_2f_1^g(x_2)\,d\hat\sigma_{gg\rightarrow gg}\,d_1^g(z ) } \label{uitdrukking}\\ & = b(y)\frac{\sum_qf_1^q(x_1)d_1^q(z)}{f_1^g(x_1)d_1^g(z ) } + b(1{-}y)\frac{\sum_qf_1^q(x_1)}{f_1^g(x_1 ) } + b(y)\frac{\sum_qf_1^q(x_2)}{f_1^g(x_2 ) } + b(1{-}y)\frac{\sum_qf_1^q(x_2)d_1^q(z)}{f_1^g(x_2)d_1^g(z)}\ .\end{aligned}\ ] ] the function @xmath140 is quite sensitive to the choice of unpolarized fragmentation functions , as demonstrated in fig .
[ akkvsfsv ] , which shows a comparison using the functions by de florian et al .
@xcite ( fsv ) and by albino et al .
@xcite ( akk ) , for a factorization scale of 10 gev .
[ lc][lc]@xmath141 ( a ) ( b ) for the akk set the assumption of gluon - gluon scattering dominance may be good to 30% , but for the fsv set one can reach this conclusion only for small @xmath40 values . at alice @xmath0 s with transverse momentum of several gev are expected to be measured in sufficiently large quantities .
a typical @xmath40 value could be @xmath142 for a 5 gev @xmath0 within a 50 gev jet .
therefore , we expect the midrapidity data to be restricted to small @xmath40 values at lhc and the assumption of gluon - gluon scattering dominance to be valid despite the large variation in @xmath0 fragmentation function sets
. a future investigation of @xmath0 production at lhc or rhic should , therefore , first focus on describing the unpolarized cross section correctly , especially regarding the @xmath108-dependence , before the @xmath108-dependence of the polarization effect is studied .
therefore , even if no polarization effect is seen , the unpolarized data obtained by the proposed measurement can in any case lead to a considerable improvement in the determination of the unpolarized @xmath0 fragmentation functions . to illustrate the main idea regarding the @xmath108 dependence of the asymmetry
, we will simply assume that the unpolarized cross section is known sufficiently well and that gluon - gluon scattering dominates , and consider two distinct scenarios : 1 ) @xmath143 ; 2 ) @xmath144 or more generally , @xmath145 .
in scenario 1 ( @xmath143 ) one finds from eq .
( [ ratio ] ) : @xmath146g}\ , \widetilde d{}_{1t}^{g\,\perp}(z{,}k_{\lambda\,t}^2 ) } { x_1f_1^g(x_1)\,x_2f_1^g(x_2)\ , d\hat\sigma_{gg\rightarrow gg}\,d_1^g(z{,}k_{\lambda\,t}^2)}\nonumber\\ & = \frac{d_{1t}^{g\,\perp}(z{,}k_{\lambda\,t}^2)}{d_1^g(z{,}k_{\lambda\,t}^2 ) } + a_g(y ) \frac{\widetilde d{}_{1t}^{g\,\perp}(z{,}k_{\lambda\,t}^2 ) } { d_1^g(z{,}k_{\lambda\,t}^2)}\ .\end{aligned}\ ] ] if there would be no wilson line or color flow dependence in the fragmentation process in the way as has been claimed in refs .
@xcite , @xmath147 would arise with the usual partonic cross section . in that case
there is no need to consider two separate functions to begin with and one should not find any @xmath108 dependence at all . observing the @xmath108 dependence @xmath148 ( which is rather weak unfortunately ) would be a signal for color flow dependence of the @xmath13-odd effects in the fragmentation process . in scenario 2 ( @xmath145 )
one finds on the other hand : @xmath149g}\ , \widetilde d{}_{1t}^{q\,\perp}(z{,}k_{\lambda\,t}^2)\,\big ) } { x_1f_1^g(x_1)\,x_2f_1^g(x_2)\ , d\hat\sigma_{gg\rightarrow gg}\,d_1^g(z{,}k_{\lambda\,t}^2)}\nonumber\\ & \mspace{80mu } + \frac{\sum_q\big(\,x_1f_1^g(x_1)\,x_2f_1^q(x_2)\ , d\hat\sigma_{gq\rightarrow
qg}\,d_{1t}^{q\,\perp}(z{,}k_{\lambda\,t}^2 ) + x_1f_1^g(x_1)\,x_2f_1^q(x_2)\,d\hat\sigma_{gq\rightarrow[q]g}\ , \widetilde d{}_{1t}^{q\,\perp}(z{,}k_{\lambda\,t}^2)\,\big ) } { x_1f_1^g(x_1)\,x_2f_1^g(x_2)\ , d\hat\sigma_{gg\rightarrow gg}\,d_1^g(z{,}k_{\lambda\,t}^2 ) } \displaybreak[0]\nonumber\\ & = b(y)\,\frac{\sum_qf_1^q(x_1)d_{1t}^{q\,\perp}(z{,}k_{\lambda\,t}^2 ) } { f_1^g(x_1)d_1^g(z{,}k_{\lambda\,t}^2 ) } + b(y)a_q(y)\ , \frac{\sum_qf_1^q(x_1)\widetilde d{}_{1t}^{q\,\perp}(z{,}k_{\lambda\,t}^2 ) } { f_1^g(x_1)d_1^g(z{,}k_{\lambda\,t}^2)}\nonumber\\ & \mspace{80mu } + b(1{-}y)\ , \frac{\sum_qf_1^q(x_2)d_{1t}^{q\,\perp}(z{,}k_{\lambda\,t}^2 ) } { f_1^g(x_2)d_1^g(z{,}k_{\lambda\,t}^2 ) } + b(1{-}y)a_q(1{-}y)\ , \frac{\sum_qf_1^q(x_2)\widetilde d{}_{1t}^{q\,\perp}(z{,}k_{\lambda\,t}^2 ) } { f_1^g(x_2)d_1^g(z{,}k_{\lambda\,t}^2)}\ .\ ] ] this expression has a rather involved @xmath108 dependence .
it is probably too difficult to disentangle the separate contributions .
therefore , we suggest a further step .
since the momentum fractions @xmath102 and @xmath103 are fixed by the jet momenta , one can select those events where they are equal , which corresponds to @xmath150 . in that case
the expression simplifies @xmath151 d_{1t}^{q\,\perp}(z{,}k_{\lambda\,t}^2 ) + \left[\,b(y)a_q(y){+}b(1{-}y)a_q(1{-}y)\,\right ] \widetilde d{}_{1t}^{q\,\perp}(z{,}k_{\lambda\,t}^2)\,\right\ } } { f_1^g(x_1)d_1^g(z{,}k_{\lambda\,t}^2)}\ , \label{ratio2}\end{aligned}\ ] ] where now @xmath152 and @xmath153 . the two different combinations of @xmath108-dependent terms are depicted in fig . [ abfuncts]b as a function of @xmath110
one can see that one term in eq .
( [ ratio2 ] ) varies more strongly with @xmath110 than the other term .
this may possibly allow a discrimination of the two effects .
in this paper we have proposed a measurement of @xmath0 polarization in the process @xmath1 , where the @xmath0 is part of one of the two observed jets which are almost back - to - back in the plane perpendicular to the beam axis . unlike the traditional measurement in the process
@xmath2 , this new observable need not vanish at midrapidity .
this makes it of interest for high energy collider experiments , such as rhic , tevatron and lhc , as they typically can detect @xmath0 s at midrapidity .
we have studied the asymmetry for the alice experiment specifically , due to the fact that alice has good pid capabilities that allow detection of @xmath0 s of momenta of several gev/@xmath154 .
the experimental situation at lhc furthermore leads to the dominance of gluon - gluon and gluon - quark scattering , allowing for a simplification of the expressions to good approximation .
observation of a nonzero @xmath112 asymmetry in the jet - jet c.o.m .
frame indicates a nonzero sivers effect in the fragmentation process .
the @xmath108 dependence of this observable in principle could allow for a study of the relative importance of the two types of interactions that could exhibit such a sivers effect .
it allows for a study of possible color - flow dependence of the asymmetry and of the ( non)universality of the sivers fragmentation functions , issues currently of much interest and that can not be addressed in @xmath2 .
but even if no asymmetry is observed , the unpolarized data that is obtained from the polarization measurement would help to constrain the unpolarized @xmath0 fragmentation functions , which are currently not well - determined .
high energy collider experiments such as to be performed at lhc would be very helpful in this respect , as the question of factorization of the unpolarized cross section would not pose a problem .
the study of color flow dependence we suggest here can be done without an actual extraction of @xmath24 , namely by just studying the dependence of the observable on particular kinematic variables .
this should prove useful even if a trustworthy extraction of the sivers fragmentation functions themselves turns out to be too difficult .
currently no knowledge on the magnitude of the various sivers fragmentation functions is available , therefore , no predictions can be given for the actual size of the asymmetry .
but the @xmath108 dependence of the various contributing terms could be derived .
they were discussed for two different scenarios : one in which the gluon sivers fragmentation function is important to include and one in which it is irrelevant compared to the quark function .
the proposed measurement allows for a determination of the @xmath40 and @xmath65 dependence of the sivers fragmentation functions . for completeness
, we mention that there are other processes from which this information can be extracted .
this can be done in electron - positron annihilation experiments via the process @xmath155 in a straightforward manner @xcite .
it allows to probe the combination @xmath156 .
it can also be done via neutral or charged current semi - inclusive dis : @xmath157 , where the combination @xmath158 is probed .
the comparison to the functions from @xmath1 would allow for another test of the ( non)universality of the sivers fragmentation functions .
we thank mauro anselmino , federico antinori , federico carminati , umberto dalesio , do - won kim , seyong kim , sungchul lee and werner vogelsang for useful discussions .
this work was supported in part by the international cooperation program of the kicos ( korea foundation for international cooperation of science & technology ) .
the work of c.b . was supported by the foundation for fundamental research of matter ( fom ) and the national organization for scientific research ( nwo ) .
a. lesnik _ et al .
_ , phys . rev
. lett .
* 35 * ( 1975 ) 770 .
g. bunce _
et al . _ , phys .
* 36 * ( 1976 ) 1113 . for a review of experimental data , see , _
e.g. _ , k. heller , in the proceedings of spin96 , edited by c.w .
de jager _
( world scientific , 1997 ) ; or a. d. panagiotou , int .
j. mod .
a * 5 * ( 1990 ) 1197 . for a review of theoretical models , see , _ e.g. _ , j. felix , mod .
a * 14 * ( 1999 ) 827 .
d. w. sivers , phys .
d * 41 * ( 1990 ) 83 ; phys .
d * 43 * ( 1991 ) 261 .
j. c. collins , nucl .
b * 396 * ( 1993 ) 161 .
m. anselmino , m. boglione and f. murgia , phys .
b * 362 * ( 1995 ) 164 .
d. boer and p. j. mulders , phys .
d * 57 * ( 1998 ) 5780 .
d. l. adams _ et al .
_ [ fnal - e704 collaboration ] , phys . lett .
b * 264 * ( 1991 ) 462 .
s. j. brodsky , d. s. hwang and i. schmidt , phys .
b * 530 * ( 2002 ) 99 .
a. airapetian _ et al . _
[ hermes collaboration ] , phys .
* 94 * ( 2005 ) 012002 .
d. boer and w. vogelsang , phys . rev .
d * 69 * ( 2004 ) 094025 .
a. bacchetta , c. j. bomhof , p. j. mulders and f. pijlman , phys .
d * 72 * ( 2005 ) 034030 .
b. i. abelev _ et al . _ [ star collaboration ] , arxiv:0705.4629 [ hep - ex ] .
p. j. mulders and r. d. tangerman , nucl .
b * 461 * ( 1996 ) 197 [ erratum - ibid .
b * 484 * ( 1997 ) 538 ] .
m. anselmino , d. boer , u. dalesio and f. murgia , phys .
d * 63 * ( 2001 ) 054029 .
j. c. collins , phys .
b * 536 * ( 2002 ) 43 .
x. ji and f. yuan , phys .
b * 543 * ( 2002 ) 66 .
a. v. belitsky , x. ji and f. yuan , nucl . phys .
b * 656 * ( 2003 ) 165 .
d. boer , p. j. mulders and f. pijlman , nucl .
b * 667 * ( 2003 ) 201 .
s. j. brodsky , d. s. hwang and i. schmidt , nucl .
b * 642 * ( 2002 ) 344 .
c. j. bomhof , p. j. mulders and f. pijlman , phys .
b * 596 * ( 2004 ) 277 .
c. j. bomhof , p. j. mulders and f. pijlman , eur .
j. * c47 * ( 2006 ) 147 - 162 .
c. j. bomhof and p. j. mulders , jhep * 0702 * ( 2007 ) 029 .
p. g. ratcliffe and o. v. teryaev , arxiv : hep - ph/0703293 .
j. collins and j. w. qiu , phys .
d * 75 * ( 2007 ) 114014 .
c. j. bomhof , p. j. mulders , w. vogelsang and f. yuan , phys .
d * 75 * ( 2007 ) 074019 .
a. metz , phys .
b * 549 * ( 2002 ) 139 .
j. c. collins and a. metz , phys .
* 93 * ( 2004 ) 252001 .
d. de florian , m. stratmann and w. vogelsang , phys .
d * 57 * ( 1998 ) 5811 .
s. albino , b. a. kniehl and g. kramer , nucl .
b * 734 * ( 2006 ) 50 .
g. l. kane , j. pumplin and w. repko , phys .
* 41 * ( 1978 ) 1689 .
m. anselmino , d. boer , u. dalesio and f. murgia , phys .
d * 65 * ( 2002 ) 114014 .
m. anselmino , m. boglione , u. dalesio , e. leader and f. murgia , phys .
d * 71 * ( 2005 ) 014002 .
a. bacchetta , c. bomhof , u. dalesio , p.
j. mulders and f. murgia , arxiv : hep - ph/0703153 .
a. bacchetta , u. dalesio , m. diehl and c. a. miller , phys .
d * 70 * ( 2004 ) 117504 .
a. bacchetta _ et al .
_ , jhep * 0702 * ( 2007 ) 093 .
d. indumathi , h. s. mani and a. rastogi , phys .
d * 58 * ( 1998 ) 094014 . c. boros , j. t. londergan and a. w. thomas , phys .
d * 62 * ( 2000 ) 014021 .
d. boer , r. jakob and p. j. mulders , nucl .
b * 504 * ( 1997 ) 345 .
|
we consider polarized @xmath0 hyperon production in proton - proton scattering , @xmath1 , in the kinematical region of the lhc experiments , in particular the alice experiment .
we present a new @xmath0 polarization observable that arises from the sivers effect in the fragmentation process . it can be large even at midrapidity and therefore , is of interest for high energy hadron collider experiments .
apart from its potential to shed light on the mechanisms behind the phenomenon of @xmath0 polarization arising in unpolarized hadronic collisions , the proposed observable in principle also allows to test the possible color flow dependence of single spin asymmetries and the ( non)universality of transverse momentum dependent fragmentation functions .
|
consider a branching random walk ( brw ) on a graph described as follows .
we are given a graph @xmath0 and a vertex @xmath1 .
we begin with one particle at @xmath2 at time @xmath3 . for each @xmath4 , each particle alive at time @xmath5 dies and gives birth to an independent random number of offspring particles according to some probability distribution @xmath6 , each of which independently takes a step according to a specified random walk on the graph with transition kernel @xmath7 .
we write @xmath8 for such a branching random walk .
we assume throughout that @xmath0 is infinite , @xmath7 is irreducible , @xmath6 has finite mean @xmath9 and @xmath10 .
we write @xmath11 for a random walk on @xmath0 with transition kernel @xmath7 , and @xmath12 for a probability measure under which the brw and the random walk are independent and begin at vertex @xmath2 .
let @xmath13 be the spectral radius of @xmath7 , @xmath14 ( which is independent of the choice of @xmath2 ) .
we say that a brw is _ recurrent _ if every vertex of the graph is visited infinitely often by the particles of the brw . otherwise it is _
transient_. benjamini and peres ( see @xcite ) gave a characterization for transience of the process in terms of the spectral radius @xmath15 of the underlying walk : they showed that if @xmath16 then the brw is transient , while if @xmath17 then the brw is recurrent .
we call a @xmath8 _ critical _ if @xmath18 . gantert and mller @xcite and bertacchi and zucca @xcite proved that any critical brw is transient .
the trace of the critical process ( the subgraph induced by edges that are traversed by particles of the brw ) is then an interesting random structure in its own right . benjamini and mller @xcite showed that on cayley graphs , simple random walk on the trace of any transient brw is itself almost surely transient , but any non - trivial _ branching _ random walk on the trace is strongly recurrent . in this work
we investigate the number of topological ends of the trace .
that is , if we remove a large ball about the origin , how many connected components do we see ? we prove the following theorem , which shows that on many graphs the trace has infinitely many ends .
we say that a random walk @xmath19 is _ quasi - symmetric _ if there exists @xmath20 such that for all vertices @xmath2 and @xmath21 , and all @xmath4 , @xmath22 in particular , simple random walk on any graph of bounded degree is quasi - symmetric .
[ vtran ] suppose that @xmath19 is quasi - symmetric , and that @xmath23 if @xmath24 , then the trace of @xmath8 has infinitely many ends almost surely . of course , there are cases which are not covered by theorem [ vtran ] .
we highlight several interesting and illuminating examples , finding cases where critical brw has one end , infinitely many ends , and everything in between .
take a random walk @xmath25 on a graph @xmath0 with transition kernel @xmath7 .
we shall associate a random walk with its transition kernel ; for example , if we say `` @xmath7 is simple random walk on @xmath0 '' we mean that @xmath7 is the transition kernel corresponding to simple random walk on @xmath0 .
recall that the spectral radius of @xmath7 is defined to be @xmath26 it is easy to see that this does not depend on the choice of vertices @xmath2 and @xmath21 , due to irreducibility of @xmath7 .
see for example @xcite for many more details on the spectral radius of a random walk .
we say that a random walk is _ amenable _ if @xmath27 , and _ non - amenable _ otherwise if and only if the graph is amenable .
this powerful result motivates our terminology . ] .
often we say that a _ graph _
@xmath0 is amenable ( non - amenable , respectively ) , by which we mean that the relevant random walk on @xmath0 is amenable ( non - amenable ) . in many cases we will consider simple ( isotropic ) random walk on a graph @xmath0 , which jumps to each of its current neighbours with equal probability , and write @xmath28 for its transition kernel .
it is natural , at first , to restrict ourselves to vertex - transitive graphs .
if a graph is amenable then the critical branching random walk is just a random walk , so we are interested in non - amenable graphs .
observe also that if a vertex - transitive graph @xmath0 has infinitely many ends itself , then the trace of any transient brw will have infinitely many ends too .
thus we are interested in one - ended , non - amenable , vertex - transitive graphs . a natural way to construct such graphs is to take the cartesian product of two vertex - transitive graphs , at least one of which is non - amenable . to be precise , for two graphs @xmath29 and @xmath30 ( not necessarily vertex transitive ) , the cartesian product @xmath31 is the graph with vertex set @xmath32 and edge set @xmath33 given transition kernels @xmath34 and @xmath35 on @xmath29 and @xmath30 respectively , we can then define a random walk on @xmath31 by flipping a fair coin at every step : if it comes up heads , then we take a step along an edge inherited from @xmath29 according to @xmath34 , and if it comes up tails , we take a random walk along an edge inherited from @xmath30 according to @xmath35 .
we write @xmath36 for the corresponding transition kernel .
we consider this to be the `` natural '' random walk on @xmath31 .
note that this is not necessarily the usual isotropic simple random walk on @xmath31 , in which the walk moves to each of its neighbours with equal probability .
however , all of our results apply without change in the isotropic case ; see section [ appendix ] for details .
our first example is @xmath38 , where @xmath39 is the tree in which every vertex has degree 3 .
consider @xmath40 .
it is easy to calculate that @xmath41 , and of course @xmath42 . from this
we deduce ( see lemma [ cs2 ] ) that @xmath43 [ t3z ] if @xmath44 , then the trace of @xmath45 has infinitely many ends almost surely .
consider the projection of the branching random walk onto @xmath39 . at each time
@xmath4 , each particle branches independently into a random number of particles with law @xmath6 ; and each of these particles moves independently according to _ lazy _ simple random walk on @xmath39 ; that is , it stays put with probability @xmath46 , otherwise it makes a step according to a simple random walk on @xmath39 . call the transition kernel for this walk @xmath47 ( so @xmath48 , where @xmath49 is simple random walk on @xmath39 and @xmath50 is the identity matrix ) .
then @xmath51 so @xmath52 is a critical branching random walk and therefore transient .
we deduce that each copy of @xmath53 is hit only finitely often by particles in @xmath45 ; since @xmath39 has infinitely many ends , this property is inherited by the trace of the branching random walk .
note that our proof did not require many detailed properties of the two graphs @xmath39 and @xmath53 .
in fact all that we used was that critical brw on @xmath39 has infinitely many ends , and that @xmath53 is amenable .
the natural next question is to consider what happens when our cartesian product is of two non - amenable graphs .
we begin with @xmath54 , and note that our previous tactic no longer works .
each copy of @xmath39 is hit infinitely often by particles of the critical brw , which tells us nothing about the number of ends of the trace .
we instead look from a different viewpoint : if we start two independent critical branching random walks from vertices a long way apart , do they meet ?
start one critical brw from a vertex @xmath55 , and call this the _ red _ process ; start the other , independent critical brw from vertex @xmath21 , and call it the _ blue _ process .
colour red any vertex that is hit by a particle in the red process , and colour blue any vertex that is hit by a particle in the blue process .
vertices that are coloured both red and blue we call _
purple_. we are interested in the number of purple vertices . [ purplefinite ] for any @xmath2 and @xmath21 , the expected number of purple vertices is finite . for this we will need some detailed estimates on the return probability of simple random walk on @xmath54 , and a simple tool for calculating expected numbers of particles in branching processes .
we delay these details for a moment to show that proposition [ purplefinite ] is enough to establish that the trace of critical brw on @xmath54 has infinitely many ends .
let @xmath7 be simple random walk on @xmath54 , and suppose that @xmath56 .
then the trace of @xmath57 has infinitely many ends almost surely . for every @xmath58 define @xmath59 to be the set of particles alive in the branching random walk at time @xmath5 .
now fix @xmath60 , and let @xmath61 , the first time that we have more than @xmath62 particles alive ( since @xmath63 and @xmath10 , @xmath64 almost surely ) .
label these particles @xmath65 .
for each particle @xmath66 at time @xmath67 , given its position , its descendants draw out a brw ; call this @xmath68 . note that , by proposition [ purplefinite ] , the intersection of the trace of @xmath68 with the trace of @xmath69 is finite for each @xmath70 .
thus the particles of @xmath68 form at least one topological end distinct from the particles of @xmath71 .
we deduce that we have at least @xmath62 ends almost surely .
since @xmath62 was arbitrary , we must have infinitely many ends .
we now list the technical results that we need to prove proposition [ purplefinite ] .
we say that @xmath72 if @xmath73 as @xmath74 through values such that @xmath75 .
the following lemma is due to cartwright and soardi ( * ? ? ?
* theorem 2i ) ) .
slight variations can also be found in @xcite and @xcite .
[ cs1 ] if @xmath67 is the homogeneous tree in which every vertex has exactly @xmath76 neighbours , and @xmath77 is a simple random walk on @xmath67 , then for any vertex @xmath2 , there exists a constant @xmath78 such that @xmath79 where @xmath15 is the spectral radius of the random walk .
combining this with another result of cartwright and soardi @xcite , included in full generality in section [ appendix ] as lemma [ cs2 ] , gives us the following corollary .
[ tp : txt ] if @xmath80 is simple random walk on @xmath54 , then for any vertex @xmath2 , there exists a constant @xmath78 such that @xmath81 where @xmath82 . finally , we will use the following well - known result about branching random walks . consider a @xmath8 .
let @xmath59 be the set of particles alive in the branching random walk at time @xmath5 , and for @xmath83 , write @xmath84 for its position in @xmath0 .
[ manytoone ] for any @xmath2,@xmath21 and @xmath5 , @xmath85 = m^n{\mathbb{p}}_i(x_n = j).\ ] ] we can now prove proposition [ purplefinite ] , which said that the expected number of purple vertices is finite when we start red and blue independent critical brws from vertices @xmath2 and @xmath21 respectively , for any choice of @xmath2 and @xmath21 in @xmath54 .
let @xmath86 be the set of particles in the red brw at time @xmath5 , and @xmath87 the set of particles in the blue brw at time @xmath5 .
then , by lemma [ manytoone ] , @xmath88 & \leq { \mathbb{e}}\left[\sum_{x\in t_3\times t_3 } \sum_{k=0}^\infty \sum_{n=0}^\infty { \mathbbm{1}}_{\{\exists u\in
n^r(k ) , v\in n^b(n ) : z(u)=z(v)=x\}}\right]\\ & \leq \sum_{x\in t_3\times t_3}\sum_{k=0}^\infty\sum_{n=0}^\infty { \mathbb{e}}[\#\{u\in n^r(k ) : z(u)=x\}]{\mathbb{e}}[\#\{v\in n^b(n ) : z(v)=x\}]\\ & \leq \sum_{x\in t_3\times t_3}\sum_{k=0}^\infty\sum_{n=0}^\infty m^k{\mathbb{p}}_i(x_k = x)\cdot m^n{\mathbb{p}}_j(x_n = x)\\ & = \sum_{k ,
n } m^{k+n}{\mathbb{p}}_{i , j}(x_k = x'_n)\end{aligned}\ ] ] where under @xmath89 , @xmath90 and @xmath91 are independent simple random walks started from @xmath2 and @xmath21 respectively .
but by the symmetric nature of our random walk , @xmath92 , so @xmath93 \leq \sum_{k , n}m^{k+n } { \mathbb{p}}_i(x_{k+n } = j).\ ] ] we now use corollary [ tp : txt ] , which tells us that @xmath94 . since @xmath95 , we obtain @xmath93 \leq c ' \sum_{k , n } \frac{1}{(k+n)^3 } < \infty.\qedhere\ ] ] having proved proposition [ purplefinite ] , and thus established that critical branching random walk on @xmath54 has infinitely many ends , we now ask ourselves what properties of the graph we used for this result , and whether we can generalise it .
in fact the only property we used , besides the obviously necessary assumption @xmath96 , was that @xmath97 for some constant @xmath20 , which is true whenever @xmath80 is quasi - symmetric .
besides , it is easy to check that @xmath98 we have therefore implicitly proved theorem [ vtran ] .
we remark here that mller @xcite obtained the same condition for ensuring that brw on a cayley graph is _ dynamically stable _ : that is , if we construct a brw and then rerandomise each random walk step at rate 1 , there is never a time at which the brw is recurrent . relaxing our assumptions even further , we obtain the following .
[ gen ] suppose that @xmath99 where @xmath100 and @xmath101 are independent random walks with transition kernel @xmath7 .
if @xmath102 , then @xmath8 has infinitely many ends almost surely .
it is natural to ask whether a converse of theorem [ vtran ] might hold .
but we already know that a full converse can not hold : simple random walk on @xmath38 satisfies @xmath103 but the corresponding critical brw has infinitely many ends .
can we , then , construct a critical brw on a non - amenable graph whose trace is one - ended ? if we allow ourselves to bias our random walk in one direction , then the answer is yes .
( without a bias , the answer is still yes , but the construction is more difficult and we save it for later . ) we return to considering @xmath104 , but this time we bias the random walk so that on @xmath53 it is more likely to move in one direction than the other . to be precise , let @xmath34 be simple random walk on @xmath39 , and @xmath105 be the random walk on @xmath53 that moves right with probability @xmath106 and left with probability @xmath107 .
let @xmath108 .
[ t3zbias ] for @xmath109 as above with @xmath110 , the trace of critical brw on @xmath37 has one end almost surely . just as in the proof that the symmetric case has infinitely many ends ,
this essentially follows from comparing the spectral radius of lazy random walk on @xmath39 with the spectral radius of @xmath109 .
indeed , just as in the proof of proposition [ t3z ] , we have @xmath111 ; but this time @xmath112 .
we deduce that each copy of @xmath53 is hit `` often enough '' that there is only one end ; the rest of the proof is concerned with making this statement precise .
we resort to the use of purple dots .
fix a copy of @xmath53 ; call it @xmath113 .
we have established that the brw hits @xmath113 infinitely often .
take a realisation of the brw , and choose two particles .
the descendants of the first particle we call _ red _ , and the descendants of the second particle we call _ blue_. any site in @xmath113 that is hit by both red and blue particles we colour purple .
we show that there are purple sites almost surely ( in fact infinitely many of them ) ; this is enough to show that the trace of the brw has only one end , since the two particles chosen were arbitrary ( and the process eventually leaves any large ball ) .
let @xmath114 be the probability that a biased rw started from a site in @xmath113 is in @xmath113 at time @xmath5 ; otherwise defined , @xmath114 is the return probability of the lazy random walk on @xmath39 at time @xmath5 . since @xmath115
, we can choose @xmath62 such that @xmath116 fix a red particle in @xmath113 . call it @xmath117 .
let @xmath118 , and for @xmath119 define @xmath120 let @xmath121 for each @xmath21 .
then @xmath122 is a galton - watson process whose birth distribution @xmath123 satisfies @xmath124 = \rho(p(p))^{-k}p_k > 1.\ ] ] let @xmath125 .
note that the probability of @xmath126 does not depend on the choice of @xmath117 ; let @xmath127 .
we choose a red particle in @xmath113 via the following algorithm : + take a red particle @xmath128 in @xmath113 .
if @xmath129 occurs , then choose @xmath128 .
for each @xmath130 , if @xmath131 does not occur , then take a red particle @xmath132 in @xmath133 .
if @xmath134 occurs , then choose @xmath132 .
since each particle s galton - watson process is independent of the others , and each has a fixed probability of surviving forever , the algorithm terminates with probability one ; then we have an infinite sequence @xmath135 of red particles in @xmath113 each of which is at distance at most @xmath62 from another ( a random walk can not move further than distance @xmath62 in time @xmath62 ) , where @xmath62 satisfies ( [ eq : pk ] ) . for any particle @xmath136 ,
let @xmath137 be the projection of that particle s position ( in @xmath37 ) onto @xmath138 .
since the process is transient , by taking a subsequence and re - ordering if necessary , we may assume that @xmath139 for each @xmath140 .
similarly we can construct an increasing sequence of blue particles @xmath141 in @xmath113 with spacing at most @xmath62 ( where @xmath62 is such that ( [ eq : pk ] ) is satisfied ) .
from these two sequences of particles , we see that there are infinitely many pairs of red and blue particles within distance @xmath142 of each other .
since the rw has a positive probability of stepping upwards @xmath142 times in a row , there exists some @xmath143 such that each pair has probability at least @xmath144 of generating a purple vertex independently of the others .
thus we must have infinitely many purple vertices , which completes the proof . on @xmath54 ,
the situation is more complicated .
we introduce a bias on one of the factors as shown in figure [ t3fig ] : a rigorous description follows . showing an isometric embedding of @xmath53 , the corresponding height function , and random walk transition probabilities .
[ t3fig],width=415 ] choose an isometric embedding @xmath145 ; that is , choose a two - sided infinite path in @xmath39 and label the vertices on that path @xmath146 .
for each vertex @xmath2 in @xmath39 , we give it a _ height _ @xmath147 .
if we view @xmath148 as labelling certain vertices in @xmath39 , then @xmath149 is the label of the closest vertex in @xmath150 minus the graph distance between @xmath2 and @xmath150 .
note that each vertex with height @xmath151 has two neighbours of heights @xmath152 and one of height @xmath153 .
now define a random walk @xmath154 on @xmath39 which jumps from a vertex with height @xmath151 to its neighbour of height @xmath153 with probability @xmath106 , and to each of its neighbours of height @xmath152 with probability @xmath155 .
let @xmath156 .
suppose that for each @xmath106 , @xmath157 .
the trace of @xmath158 has infinitely many ends almost surely . on the other hand , if @xmath159 , then the trace of @xmath160 has one end almost surely .
the first statement follows from exactly the same argument as proposition [ t3z ] .
the second is almost identical to proposition [ t3zbias ] , except that we have to look at a copy of @xmath53 embedded in @xmath39 . *
open problem .
* how many ends does the trace of @xmath161 have when @xmath162 ?
what if we insist that our random walk is isotropic ?
can we construct a graph on which the corresponding critical brw is one - ended ?
again , the answer is yes .
let @xmath163 , and let @xmath164 be the rooted tree in which every vertex has four children , and every vertex except the root has one parent .
we say that the root has generation @xmath3 , its children have generation @xmath165 , and so on .
construct a graph @xmath166 by joining @xmath167 to every vertex in generation @xmath5 and every vertex in generation @xmath168 in @xmath164 , for each @xmath4 .
we call @xmath166 the hammock graph .
we have @xmath169 , but if @xmath170 , then @xmath171 has one end almost surely .
let @xmath172 be the vertices in generation @xmath62 of @xmath164 , together with vertex @xmath173 .
then @xmath174 and @xmath175 by coupling with a random walk on @xmath53 that jumps right with probability @xmath176 and left with probability @xmath177 , we see that @xmath178 .
starting at the root @xmath179 of @xmath164 , by repeatedly jumping to generation @xmath165 and then back to the root , we get @xmath180 so @xmath181 . showing that the critical brw is one - ended can be done in a very similar way to the proof of proposition [ t3zbias ] .
choose any two particles in the brw , and colour their descendants red and blue respectively ; any site hit by both a red and a blue particle is coloured purple . clearly
a random walk on @xmath166 eventually hits @xmath182 almost surely , so @xmath182 is hit infinitely often by red particles and infinitely often by blue particles .
we can ensure ( for example by constructing embedded galton - watson processes as in the proof of proposition [ t3zbias ] ) that there are almost surely infinitely many red - blue pairs within distance at most @xmath62 of each other .
each of these pairs has at least a fixed probability @xmath183 of creating a purple dot ( since this time @xmath182 does not have bounded degree , we have to allow particles to bounce back and forth between @xmath182 and @xmath164 ) .
thus we have infinitely many purple dots almost surely , and since our initial choice of red and blue particles was arbitrary , this guarantees that we have only one end . by gluing together various copies of graphs
already constructed , it is easy to construct graphs on which critical brw has any number of ends .
for example , if we glue two copies of @xmath166 and one copy of @xmath54 at a single vertex , then we have a graph on which critical brw has one end , two ends , or infinitely many ends , each with positive probability .
* open problem .
* does there exist a quasi - symmetric random walk such that the corresponding critical brw is one - ended almost surely ?
another possibility for future research is to investigate the cayley graphs of co - compact fuchsian groups .
these graphs correspond to tilings of the hyperbolic plane .
lalley and sellke @xcite showed that critical branching brownian motion on the hyperbolic plane has a limit set that does not have full measure , and by analogy one might expect that critical branching random walks on co - compact fuchsian groups have infinitely many ends . finally , we relay a conjecture from itai benjamini . on any vertex - transitive graph @xmath0 , the trace of simple critical brw has infinitely many ends .
as we mentioned briefly in section [ notation ] , given random walks on @xmath184 graphs @xmath185 , there is more than one natural random walk on @xmath186 . rather than overcomplicate matters with general definitions in the earlier sections , we concentrated on one such natural choice .
here we give more details , and see that in fact our results hold regardless of the choice .
given random walks @xmath187 on @xmath185 respectively , and @xmath188 with @xmath189 , we define the @xmath190-_product random walk _
@xmath90 on @xmath191 by setting @xmath192 we write @xmath193 for the transition kernel of this random walk . for vertex - transitive graphs , there are at least two natural choices for @xmath194 : we could take @xmath195 for each @xmath2 , or @xmath196 where @xmath197 is the degree of an arbitrary vertex in @xmath198 .
the latter choice corresponds to simple isotropic random walk on @xmath186 .
the following lemma , due to cartwright and soardi @xcite , tells us how to translate results about return probabilities on certain graphs into results about return probabilities on their cartesian products .
we recall that the cayley graph @xmath199 of a group @xmath200 with generating set @xmath201 has as its vertex set the elements of @xmath200 , with two vertices @xmath202 joined by an edge if @xmath203 for some @xmath204 .
we write that @xmath205 if @xmath206 as @xmath74 through values such that @xmath75 . [ cs2 ] suppose that @xmath207 are cayley graphs with associated random walks @xmath187 .
let @xmath90 be the @xmath190-product random walk on @xmath191 .
fix @xmath208 .
suppose that there exist constants @xmath209 and @xmath210 such that for each @xmath211 , @xmath212 then there exists @xmath20 such that @xmath213 note in particular that the choice of @xmath214 affects the spectral radius in the obvious way , and has no effect on the polynomial terms .
thus the proofs in earlier sections are unaffected by choosing different @xmath214 .
for example , the results on @xmath215 where we used @xmath216 hold also for @xmath217 , which corresponds to simple isotropic random walk on @xmath38 .
we would like to thank itai benjamini for asking us these questions , and for several helpful discussions .
we are also grateful to steven lalley , yuval peres and yehuda pinchover . during this project
ec received support from the austrian academy of science ( doc - fforte fellowship , project number d-1503000014 ) ; microsoft research ; austrian science fund ( fwf ) : s09606 , part of the austrian national research network `` analytic combinatorics and probabilistic number theory '' ; and project dk plus , funded by fwf , austrian science fund ( project number e-1503w01230 ) .
mr was supported by a crm - ism postdoctoral fellowship , mcgill university , the university of warwick , and epsrc fellowship
ep / k007440/1 .
|
we investigate the number of topological ends of the trace of branching random walk ( brw ) on a graph , showing that in many symmetric cases there are infinitely many ends .
we then describe some brws which have just one end , and conclude with some open problems .
|
multiloop scattering amplitudes in maximally supersymmetric ( @xmath0 ) yang - mills theory have been studied extensively over the years @xcite in connection with for instance the famous ads / cft correspondence and possible finiteness of supergravity theories .
remarkable results have been uncovered including new favorable evaluation methods applicable to both tree- and loop - level amplitudes .
an essential part of this progress is the on - shell superspace formalism , which organizes on - shell states and scattering amplitudes in maximally supersymmetric super yang - mills theory very elegantly @xcite .
the principle is to arrange the entire supermultiplet as a convenient expansion labeled by @xmath1-symmetry indices and particle number into @xmath2 superfields , one for each external leg .
all possible scattering combinations are realized by formation of superamplitudes , defined as generating functions with the superfields as input , having all supersymmetric ward identities automatically satisfied .
individual scattering amplitudes are available from the generating function using appropriate combinations of grassmann differential operators . using either the maximally helicity violating ( mhv ) vertex expansion @xcite or the britto - cachazo - feng - witten ( bcfw ) on - shell recursion relations @xcite superamplitudes for general particle and helicity configurations may be constructed . in a recent paper @xcite , overlapping with @xcite , the maximally supersymmetric superspace setup was generalized to super yang - mills theory with reduced supersymmetry , i.e. with @xmath3 generators of supersymmetry .
all necessary steps towards developing both a holomorphic and a non - holomorphic approach were taken . in particular
, it was shown that the most general mhv generating function valid for generic supersymmetry can be derived by simple combinations of truncations and fourier transforms of the full @xmath0 superamplitude .
the introduction of @xmath0 superamplitudes spawned important developments including a very convenient diagrammatic representation @xcite .
for brevity , all possible contractions between external states are tracked , yielding a one - to - one correspondence between diagrams and individual scattering amplitudes .
motivated by @xcite we will extend this graphical framework to @xmath3 superamplitudes .
before developing the superspace formalism and discussing the mhv and @xmath4 generating functions for super yang - mills theory with less than maximal supersymmetry , we briefly remind ourselves about the @xmath0 setup .
the @xmath0 vector multiplet is uniquely cpt self - conjugate allowing all on - shell states to be incorporated into a single holomorphic superfield @xmath5 written as an expansion in grassmann variables @xmath6 with @xmath7 being @xmath1-symmetry indices . within this framework
the sixteen physical states in the @xmath0 supermultiplet are two gluons @xmath8 and @xmath9 , four gluino pairs @xmath10 and @xmath11 , plus six real scalars @xmath12 , all completely antisymmetric .
the superfield then takes the form @xcite @xmath13 grassmann fourier transformation yields an antiholomorphic superfield , @xmath14 but with the exact same particle content encoded . because of this equivalence either representation may be preferred
in order to shed light on how proliferation of amplitudes in @xmath0 super yang - mills theory is handled , the concept of superamplitudes is introduced . with @xmath2 copies of the superfield @xmath15 arranged , we organize the full @xmath2-point tree - level superamplitude ascendingly according to grassmann degree in steps of four , @xmath16 ranging from eight @xmath17 s to @xmath18 .
explicit formulas for all @xmath19 amplitudes relying on bcfw shifts @xcite exist in the litterature @xcite , but our focus is on @xmath20 and @xmath4 amplitudes .
it follows that all mhv amplitudes may be collected into a generating function , which we will call the mhv superamplitude , such that each term corresponds to a regular scattering amplitude involving gluons , fermions and scalars .
the mhv superamplitude is defined as @xcite @xmath21 and contains in addition to the well - known overall momentum conservation , an eightfold grassmann delta function , which conserves supermomentum @xmath22 .
it proves advantageous to expand the superamplitude as a sum of monomials in the @xmath17 s .
factorization in the group index and @xmath23 for grassmann variables imply that @xmath24 consequently , the mhv superamplitude can be recast as @xmath25 with four - momentum conservation stripped .
notation of component amplitudes is streamlined in terms of spinor products of supermomenta of the individual legs defined by @xmath26}\equiv \tilde\eta_i^a{[ij]}\tilde\eta_j^a\;.\end{aligned}\ ] ] for four external legs , some simple examples of component amplitudes are @xmath27 analogous to the mhv superamplitude one can define the @xmath4 superamplitude @xcite by @xmath28 } } = i(-1)^n\frac{\prod_{a=1}^4\sum_{i < j}^n{[\tilde q_i^a\tilde q_j^a ] } } { \prod_{m=1}^{n}{[m(m+1)]}}\;,\end{aligned}\ ] ] built entirely from antiholomorphic superfields
. it is mapped from the @xmath29 coordinates to the untilded superspace using the grassmann fourier transform .
it is desirable to extend the neat @xmath0 superspace formulation to super yang - mills theory with nonmaximal supersymmetry .
two different approaches exist @xcite , neither of which can be described in terms of a single superfield , due to the fact that @xmath3 super yang - mills theory is not cpt self - conjugate .
instead the physical states of the @xmath30 supermultiplets can be assembled into either two conjugate superfields @xmath31 and @xmath32 , which are related to and by the truncations @xmath33 and @xmath34 respectively , or two holomorphic superfields @xmath31 and @xmath35 obtained from the @xmath0 superfield by suitable combinations of truncations and grassmann integrations . here
we focus on the @xmath36 formalism , since we would like to avoid the unfortunate mixing of the @xmath17 and @xmath29 variables in the @xmath37 picture , which does not lead to any obvious graphical interpretation .
consider , say , the @xmath38 conjugate superfields obtained by letting @xmath39 and @xmath40 in and .
it follows that @xmath41 and @xmath42 .
the @xmath2-point mhv configuration then has two states from the @xmath43 superfield and @xmath44 from the other .
the essense of the @xmath36 approach is to keep the truncated holomorphic superfields , but discard the ones of negative overall helicity . to achieve this
we observe that full description of the particle content can be maintained by supplementing @xmath15 by a new holomorphic superfield introduced in @xcite , @xmath45 where for instance @xmath46 with @xmath47 fixed but kept for notational uniformity .
for clarity we list here all pairs of superfields @xmath48 $ ] for @xmath30 .
states of the @xmath49-sector are hatted to be distinguished from those originating from the non - manipulated superfields . in this notation
the indices take values @xmath50 .
@xmath51 let us define what is understood by a mhv amplitude in @xmath3 super yang - mills theory in this notation .
first of all , a mhv amplitude must have @xmath52 paired indices as a consequence of truncation of the superfields . moreover , to be mhv requires two states from the @xmath49 sector and @xmath44 from the @xmath53 superfield .
with this in mind we are now ready to derive the mhv superamplitude for @xmath3 .
recall that the maximally supersymmetric superamplitude is a function of @xmath15 , @xmath54
. suppose legs @xmath55 and @xmath56 represent states of the @xmath49-sector .
we can then convert @xmath15 and @xmath57 to @xmath58 and @xmath59 by integrating out @xmath60 @xmath61 s and @xmath62 s according to .
afterwards we truncate the remaining @xmath44 legs to reduce the content of the @xmath53 superfields .
the only obstacle is to rearrange the integration measure appropriately , but actually the overall sign is not very important to us . @xmath63
it is noticed that this generic mhv superamplitude , which is equivalent to the one derived in @xcite , as expected reduces to the original @xmath0 mhv superamplitude for maximal supersymmetry , and to the parke - taylor formula for pure yang - mills theory ( @xmath64 ) .
it would be useful to have a simple visualization of the rather abstract expression for the @xmath65 superamplitude .
similar to feynman graphs , such a scheme should yield a one - to - one correspondence between diagrams and component amplitudes .
the basic ingredients of the @xmath0 mhv superamplitude appear in as a cyclic spinor string in the denominator and spinor products of supermomenta in the numerator .
the pictorial representation for these components was developed in @xcite and the transition to @xmath3 was sketched . taking into account
we see that the only structural difference between the maximally supersymmetric and the @xmath3 generating function is the number of spinor products of supermomenta .
contrary to our @xmath3 mhv superamplitude , the @xmath66 integrated index lines responsible for the @xmath67 factor still carry grassmann variables with fixed @xmath1-symmetry indices in @xcite .
we circumvent these by introducing a new _ sector line _ between the two states of the @xmath49-sector to catch the @xmath67 factor of .
the individual terms in the spinor string in the denominator are diagrammatrically represented by a curved line without endpoints .
the blue index line with endpoints translates into a spinor product of supermomenta of the corresponding individual legs . for @xmath3 , states of the @xmath49-sector
must be connected with a solid green sector line without end points .
identical graphs exist in the @xmath4 picture with obvious continued expressions . ] in order to construct an index diagram for a given mhv amplitude with @xmath2 external legs in @xmath65 super yang - mills theory having @xmath52 paired @xmath1-symmetry indices , simply follow this prescription : 1 .
draw a polygon with @xmath2 sides of solid , black lines curving inwards , leaving space at each corner for external legs .
2 . distribute the external legs at the gaps and label them with appropriate momentum , helicity and @xmath1-symmetry indices .
3 . connect paired @xmath1-symmetry indices with blue index lines with endpoints .
4 . for @xmath3 insert a single solid green sector line without endpoints between the two states of the @xmath49-sector .
5 . indicate holomorphicity with @xmath68 for mhv and @xmath69 for @xmath4 in the center of the diagram .
the diagram rules are summarized in fig .
[ indexrules ] .
let us now see how the diagrammatic representation scheme works out in practice for a number of mhv amplitudes at tree - level in super yang - mills theory with @xmath0 and fewer supersymmetries .
we start with the amplitudes expressions - , whose corresponding diagrams are drawn fig .
[ n=4indexdiagrams ] .
these amplitudes are viewed for maximal supersymmetry without sector lines , but actually they exist for all @xmath70 , @xmath71 and @xmath72 respectively . in fig . [ n=2indexdiagrams ] the corresponding @xmath73 graphs are given .
thanks to the @xmath74 factor from the sector line the diagrams of figs .
[ n=4indexdiagrams ] and [ n=2indexdiagrams ] have pairwise identical expressions .
the analytic expressions for the @xmath0 amplitudes - are neatly captured by these three simple index diagrams .
if reinterpreted in the @xmath4 picture the first diagram would have four index lines between the two positive helicity gluons for instance.,title="fig : " ] the analytic expressions for the @xmath0 amplitudes - are neatly captured by these three simple index diagrams .
if reinterpreted in the @xmath4 picture the first diagram would have four index lines between the two positive helicity gluons for instance.,title="fig : " ] the analytic expressions for the @xmath0 amplitudes - are neatly captured by these three simple index diagrams .
if reinterpreted in the @xmath4 picture the first diagram would have four index lines between the two positive helicity gluons for instance.,title="fig : " ] many amplitudes exist for several values of @xmath70 . here - are shown for @xmath73 . in general , if a diagram has at most @xmath75 grouped index lines in @xmath0 , then it can be non - zero for reduced supersymmetry only provided @xmath76.,title="fig : " ] many amplitudes exist for several values of @xmath70 . here - are shown for @xmath73 . in general , if a diagram has at most @xmath75 grouped index lines in @xmath0 , then it can be non - zero for reduced supersymmetry only provided @xmath76.,title="fig : " ] many amplitudes exist for several values of @xmath70 . here - are shown for @xmath73 .
in general , if a diagram has at most @xmath75 grouped index lines in @xmath0 , then it can be non - zero for reduced supersymmetry only provided @xmath76.,title="fig : " ] we have now shown how easy it is to draw tree - level diagrams for mhv configurations , but it still remains to demonstrate the translation of diagrams into the matching analytic expressions .
consider , say , the six - point @xmath77 and a seven - point @xmath73 mhv tree - level amplitudes of fig .
[ sixsevenpoint ] .
these diagrams were constructed simply by selecting two states from the @xmath49 superfield and then patching up using @xmath53 respecting that the total number of indices should be @xmath52 . referring to fig .
[ indexrules ] we almost effortlessly find @xmath78 @xmath77 and @xmath73 mhv tree - level examples as suggested by the number of index lines .
, title="fig : " ] @xmath77 and @xmath73 mhv tree - level examples as suggested by the number of index lines . ,
title="fig : " ]
despite being formulated for tree - level amplitudes , the idea of generating functions and their diagrammatic representation fits excellently with evaluation of loop amplitudes using the generalized unitarity cut method @xcite .
indeed , the required supersymmetric sum over all possible on - shell states propagating in the intermediate channels may be realized as grassmann integration of superamplitudes , while the flow of @xmath1-symmetry charges is captured by index diagrams . as an example we study a one - loop unitarity cut of a four - point amplitude with external gluons , and carry out the supersum for the @xmath73 case . for a more thorough description of unitarity cuts see @xcite and later developments @xcite .
evaluation of supersums has been discussed in @xcite . here
we follow the strategy of @xcite .
the supersum in question receives in total eight contributions , two with internal gluons , four with a fermion loop and two having scalars .
[ loopdiagrams ] provides two of these diagrams , and using the rules given in fig .
[ indexrules ] , we find that their numerator part translate into @xmath79 } { \langleq_{2b}q_{\ell_1 b}\rangle}{[\tilde{q}_{\ell_1}^b\tilde{q}_4^b ] } { \langle2|\ell_1|4]}^2 \quad \mbox{and } \quad { \langleq_{2a}q_{\ell_1 a}\rangle}{[\tilde{q}_{\ell_1}^a\tilde{q}_4^a ] } { \langleq_{2b}q_{\ell_2 b}\rangle}{[\tilde{q}_{\ell_2}^b\tilde{q}_4^b ] } { \langle2|\ell_1|4]}^2,\end{aligned}\ ] ] where @xmath80 } = { \langlei|j|k]}$ ] for shorthand .
the left and right index diagrams should respectively illustrate internal gluon and fermion contributions in a unitarity cut of the four - point one - loop amplitude .
the cut marked by the dashed red line splits the amplitude into @xmath20 and @xmath4 parts .
again the green sector line accounts for reduced supersymmetry .
horizontal flips of these diagrams and two additional diagrams representing internal scalars are very easy to draw , but are left out here .
, title="fig : " ] the left and right index diagrams should respectively illustrate internal gluon and fermion contributions in a unitarity cut of the four - point one - loop amplitude .
the cut marked by the dashed red line splits the amplitude into @xmath20 and @xmath4 parts .
again the green sector line accounts for reduced supersymmetry .
horizontal flips of these diagrams and two additional diagrams representing internal scalars are very easy to draw , but are left out here .
, title="fig : " ] by inspection all eight diagrams have the same relative sign because the corresponding grassmann expressions are ordered equivalently .
therefore the @xmath17 s may just be suppressed , and the numerator of the supersum thus becomes @xmath81}^4+{\langle2|\ell_2|4]}^4 + 2{\langle2|\ell_1|4]}^3{\langle2|\ell_2|4]}+ 2{\langle2|\ell_1|4]}{\langle2|\ell_2|4]}^3 + 2{\langle2|\ell_1|4]}^2{\langle2|\ell_2|4]}^2 \nonumber \\ = ( { \langle2|\ell_1|4]}+{\langle2|\ell_2|4]})^2 \times({\langle2|\ell_1|4]}^2+{\langle2|\ell_2|4]}^2 ) \;.\end{aligned}\ ] ] an interesting pattern emerges when comparing the supersums for @xmath82 . adding the diagrams in the right combinations dictated by the different supermultiplets we see that the displayed resummation property is a common feature .
indeed , for maximal supersymmetry the supersum takes the very compact form @xmath83}+{\langle2|\ell_2|4]})^4 $ ] , while for @xmath3 @xmath84}+{\langle2|\ell_2|4]})^{{\mathcal{n } } } \times({\langle2|\ell_1|4]}^{4-{\mathcal{n}}}+{\langle2|\ell_2|4]}^{4-{\mathcal{n } } } ) \;,\end{aligned}\ ] ] in agreement with @xcite .
cuts and supersums are by no means limited to mhv amplitudes . non - mhv amplitudes may be generated from mhv ones using the mhv vertex construction @xcite as addressed in @xcite , and the mhv techniques thus apply .
in this paper we have investigated the superspace formalism of @xmath0 super yang - mills theory and its extension to situations with less than maximal supersymmetry .
we have written a general @xmath70-dependent form of the mhv generating function at tree - level .
more importantly , an extension to @xmath70-fold supersymmetry of a recent scheme for representing @xmath0 superamplitudes diagrammatically was presented . although simple , it is nice to see that the technique carries over from @xmath0 . with this diagrammatic prescription
it is extremely easy to memorize any super yang - mills scattering amplitude and translate it into analytic expressions , as we illustrated through several examples at both tree - level and one - loop .
we will leave the exploration of multiloop unitarity cuts of non - mhv amplitudes utilizing the full content of the various @xmath3 multiplets for external states for future work .
another direction could be to study diagrams and supersums in super yang - mills theory coupled to matter .
99 z. bern , j. j. m. carrasco , h. ita , h. johansson and r. roiban , phys .
d * 80 * , 065029 ( 2009 ) [ arxiv:0903.5348 [ hep - th ] ] .
h. elvang , y. t. huang and c. peng , arxiv:1102.4843 [ hep - th ] .
z. bern , m. czakon , l. j. dixon , d. a. kosower and v. a. smirnov , phys .
d * 75 * , 085010 ( 2007 ) [ arxiv : hep - th/0610248 ] . c. anastasiou , z. bern , l. j. dixon and d. a. kosower , phys .
* 91 * , 251602 ( 2003 ) [ arxiv : hep - th/0309040 ] .
z. bern , l. j. dixon and v. a. smirnov , phys .
d * 72 * , 085001 ( 2005 ) [ arxiv : hep - th/0505205 ] .
f. cachazo , m. spradlin and a. volovich , phys .
d * 74 * , 045020 ( 2006 ) [ arxiv : hep - th/0602228 ] .
z. bern , m. czakon , d. a. kosower , r. roiban and v. a. smirnov , phys .
lett . * 97 * , 181601 ( 2006 ) [ arxiv : hep - th/0604074 ] . j. m. drummond , j. henn , g. p. korchemsky and e. sokatchev , nucl . phys .
b * 826 * , 337 ( 2010 ) [ arxiv:0712.1223 [ hep - th ] ] .
j. m. drummond , j. henn , v. a. smirnov and e. sokatchev , jhep * 0701 * , 064 ( 2007 ) [ arxiv : hep - th/0607160 ] .
korchemsky , j. m. drummond and e. sokatchev , nucl .
b * 795 * , 385 ( 2008 ) [ arxiv:0707.0243 [ hep - th ] ] .
a. brandhuber , p. heslop and g. travaglini , nucl .
b * 794 * , 231 ( 2008 ) [ arxiv:0707.1153 [ hep - th ] ] .
j. m. drummond , j. henn , g. p. korchemsky and e. sokatchev , nucl .
b * 795 * , 52 ( 2008 ) [ arxiv:0709.2368 [ hep - th ] ] .
z. bern , j. j. m. carrasco , h. johansson and d. a. kosower , phys . rev .
d * 76 * , 125020 ( 2007 ) [ arxiv:0705.1864 [ hep - th ] ] .
j. m. drummond , j. m. henn and j. plefka , jhep * 0905 * , 046 ( 2009 ) [ arxiv:0902.2987 [ hep - th ] ] .
n. berkovits and j. maldacena , jhep * 0809 * , 062 ( 2008 ) [ arxiv:0807.3196 [ hep - th ] ] . n. beisert , r. ricci , a. a. tseytlin and m. wolf , phys .
d * 78 * , 126004 ( 2008 ) [ arxiv:0807.3228 [ hep - th ] ] .
n. beisert , fortsch .
phys . * 57 * , 329 ( 2009 ) [ arxiv:0903.0609 [ hep - th ] ] .
j. bartels , l. n. lipatov and a. sabio vera , phys .
d * 80 * , 045002 ( 2009 ) [ arxiv:0802.2065 [ hep - th ] ] .
r. c. brower , h. nastase , h. j. schnitzer and c. i. tan , nucl .
b * 822 * , 301 ( 2009 ) [ arxiv:0809.1632 [ hep - th ] ]
. v. del duca , c. duhr and e. w. n. glover , jhep * 0812 * , 097 ( 2008 ) [ arxiv:0809.1822 [ hep - th ] ] .
z. bern , l. j. dixon , d. a. kosower , r. roiban , m. spradlin , c. vergu and a. volovich , phys .
d * 78 * , 045007 ( 2008 ) [ arxiv:0803.1465 [ hep - th ] ] .
v. p. nair , phys .
b * 214 * , 215 ( 1988 ) .
m. bianchi , h. elvang and d. z. freedman , jhep * 0809 * , 063 ( 2008 ) [ arxiv:0805.0757 [ hep - th ] ] .
h. elvang , d. z. freedman and m. kiermaier , jhep * 0904 * , 009 ( 2009 ) [ arxiv:0808.1720 [ hep - th ] ] . n. arkani - hamed , f. cachazo and j. kaplan , jhep * 1009 * , 016 ( 2010 ) [ arxiv:0808.1446 [ hep - th ] ] . j. m. drummond , j. henn , g. p. korchemsky and e. sokatchev , arxiv:0808.0491 [ hep - th ] .
j. m. drummond , j. henn , g. p. korchemsky and e. sokatchev , nucl .
b * 828 * , 317 ( 2010 ) [ arxiv:0807.1095 [ hep - th ] ] .
a. brandhuber , p. heslop and g. travaglini , phys .
d * 78 * , 125005 ( 2008 ) [ arxiv:0807.4097 [ hep - th ] ] .
j. m. drummond and j. m. henn , jhep * 0904 * , 018 ( 2009 ) [ arxiv:0808.2475 [ hep - th ] ] . g. georgiou , e. w. n. glover and v. v. khoze , jhep * 0407 * , 048 ( 2004 ) [ arxiv : hep - th/0407027 ] . y. t. huang , phys . lett .
b * 631 * , 177 ( 2005 ) [ arxiv : hep - th/0507117 ] .
h. feng and y. t. huang , jhep * 0904 * , 047 ( 2009 ) [ arxiv : hep - th/0611164 ] .
h. elvang , d. z. freedman and m. kiermaier , jhep * 0906 * , 068 ( 2009 ) [ arxiv:0811.3624 [ hep - th ] ] .
f. cachazo , p. svrcek and e. witten , jhep * 0409 * , 006 ( 2004 ) [ arxiv : hep - th/0403047 ]
. k. risager , jhep * 0512 * , 003 ( 2005 ) [ arxiv : hep - th/0508206 ] .
r. britto , f. cachazo and b. feng , nucl .
b * 715 * , 499 ( 2005 ) [ arxiv : hep - th/0412308 ] .
r. britto , f. cachazo , b. feng and e. witten , phys .
lett . * 94 * , 181602 ( 2005 ) [ arxiv : hep - th/0501052 ] .
s. lal and s. raju , phys .
d * 81 * , 105002 ( 2010 ) [ arxiv:0910.0930 [ hep - th ] ] .
z. bern , l. j. dixon , d. c. dunbar and d. a. kosower , nucl .
b * 435 * , 59 ( 1995 ) [ arxiv : hep - ph/9409265 ] .
z. bern , l. j. dixon , d. c. dunbar and d. a. kosower , nucl .
b * 425 * , 217 ( 1994 ) [ arxiv : hep - ph/9403226 ] .
z. bern , l. j. dixon and d. a. kosower , nucl .
b * 513 * , 3 ( 1998 ) [ arxiv : hep - ph/9708239 ] .
z. bern , v. del duca , l. j. dixon and d. a. kosower , phys .
d * 71 * , 045006 ( 2005 ) [ arxiv : hep - th/0410224 ] .
z. bern , l. j. dixon and d. a. kosower , jhep * 0408 * , 012 ( 2004 ) [ arxiv : hep - ph/0404293 ] .
r. britto , f. cachazo and b. feng , nucl .
b * 725 * , 275 ( 2005 ) [ arxiv : hep - th/0412103 ] .
|
we present an on - shell graphical framework for superamplitudes in super yang - mills theory with arbitrary supersymmetry .
our diagrammatic procedure is derived through manipulations of the full @xmath0 superamplitude and illustrated by a number of explicit examples .
|
quantum entanglement is one of the most important phenomenon in quantum theory .
it exhibit the nature of nonlocal correlation between quantum systems , and plays an essential role in various fields of quantum information processing , such as quantum computation , quantum communication , quantum cryptography , quantum teleportation@xcite , and closely related to quantum [email protected] the first experiments @xcite on quantum teleportation using two - mode squeezed states @xcite@xcite , a significant amount of work has been devoted to develop a quantum information theory of continuous variable systems .
so far , most of the theoretical work has focused on the entanglement properties of the quantum states involved in all these experiments , the so - called gaussian states .
the first problem arisen is that if a given quantum gaussian state is entangled , the problem of qualifying entanglement has been solved in the general bipartite setting @xcite@xcite .
but the efficiency of entanglement manipulation protocols used in practical quantum information processing critically depends on the quality of the entanglement that one can generate .
it is therefore essential to be able to quantify the amount of entanglement in systems with continuous variables especially gaussian states .
several measures have been proposed to quantify the amount of entanglement @xcite .
among which are the entanglement of formation , relative entropy of entanglement and distillation entanglement .
entanglement of formation is one of the most important entanglement measures . for symmetric guassian state ,
this entanglement measure can be carried out analytically @xcite . for general bipartite
gaussian state , another entanglement measure called gaussian entanglement of formation was introduced@xcite .
clearly it is the up bound of the entanglement of formation .
relative entropy of entanglement@xcite @xcite on the other hand measures the distance between the state under consideration to the closest separable state .
it has the advantage that separable states obviously correspond to zero distance . for pure states of bipartite systems
it reduces to the von neumann entropy of the reduced state of either subsystem . for mixed bipartite states it is usually difficult
to be calculated , except for some specific states .
the distance of the gaussian state to the set of separable gaussian states measured by the relative entropy was considered @xcite .
although one has yet no proof that there does not exist a non - gaussian separable state which is closer to the gaussian state under consideration than the closest separable gaussian state , one has good reason to think that it is a good bound of entanglement measure .
however , the expression derived @xcite is still not easy for a direct numeric calculation .
recently the bound on the relative entropy of entanglement was calculated for two modes squeezed symmetric thermal state@xcite . in this paper
i will consider the more practical used state of two mode squeezed non - symmetric themal state , which is to be distributed between two parties by means of a lossy optical fiber in asymmetric settings for an initial two mode squeezed vacuum [email protected] will concentrate on its gaussian entanglement of formation and its up bound of relative entropy of entanglement . according to the hypothesis of hashing inequality@xcite
, all this entanglement measures should be lower bounded by coherent information .
i will at last compare the results with coherent information of the state .
gaussian state is completely specified by its mean and its covariance matrix , where the mean can be dropped by local unitary operation so that is irrelevant for entanglement problems .
let us consider a quantum gaussian state @xmath0 of two system a and b acting on a hilbert space @xmath1 .
( in quantum optical term , it is called two modes . )
it is convenient to describe gaussian state density operator @xmath0 by its characteristic function .
( e.g. @xcite)@xmath2here @xmath3 is a real vector and @xmath4 $ ] is _ weyl operator _ ( displacement operator ) with @xmath5 and @xmath6 are operators of system a and b respectively , satisfying canonical commutation relations .
a characteristic function @xmath7 uniquely defines a state @xmath8 .
gaussian state is state
whose @xmath7 is gaussian function of @xmath9 : @xmath10 $ ] , where @xmath11 is a real symmetric matrix , called correlation matrix .
any gaussian state of two modes can be transformed into what we called the standard form , using local unitary operation only @xcite @xcite .
the corresponding characteristic function has displacement @xmath12 and the correlation matrix has the simple form of four parameters .
it was proved that ( see @xcite and reference therein , also see @xcite and reference therein ) : for arbitrary real symmetric matrix @xmath11 there exists linear transformation ( symplectic transformation ) @xmath13 , preserving canonical commutation relations , then @xmath14 this is to say that by proper symplectic transformation , gaussian state can be transformed into a direct product of thermal states @xmath15 . the one mode thermal state density operators are of the forms @xmath16 where @xmath17 , @xmath18 , @xmath19 .
the symplectic transformation @xmath20 induces a unitary operator @xmath21 so that @xmath22 @xcite@xcite , and @xmath23 . for @xmath24 is a direct product of thermal states .
it represents the randomness side of state @xmath0 . while the quantum correlation side of @xmath0 should be caused by the two mode unitary operators @xmath21 . in this paper
i will consider one of the special kind of gaussian states which can be generated from the thermal states @xmath25 with a simple form of @xmath26 .
@xmath27 \oplus \left [ \begin{array}{ll } \cosh r & \sinh r \\ \sinh r & \cosh r \end{array } \right ] , \ ] ] then @xmath28 .
the @xmath29 induced unitary operator @xmath30 @xcite is just the two - mode squeezed operator @xmath31 @xmath32 \nonumber \\ & & \cdot \exp \left ( -a_{a}a_{b}\tanh r\right ) .\end{aligned}\ ] ] this kind of gaussian states will be called squeezed non - symmetric thermal states . and in the rest of this paper , i will use @xmath0 to specify the states .
@xmath33 denote @xmath34 , then the inseparability criterion reads @xmath35 for the convenience of further application , one can express @xmath0 in coherent state representation as @xmath36 \nonumber \\ & & \cdot \exp [ \tau \left ( \alpha _ { a}^{\ast } \alpha _ { b}^{\ast } + \beta _ { a}\beta _ { b}\right ) + \omega _ { a}\alpha _ { a}^{\ast } \beta _ { a}+\omega _ { b}\alpha _ { b}^{\ast } \beta _ { b}]\end{aligned}\ ] ] where @xmath37 , @xmath38 , @xmath39 , @xmath40 . with coherent state representation
the reduced state can be easily obtained by integration .
the reduced density matrix @xmath41 and @xmath42 turn out to be one mode thermal states @xmath43 , @xmath44 with parameters @xmath45 and @xmath46 respectively , @xmath47
entanglement of formation is defined as an infimum @xmath48 over all ( possibly continuous ) convex decompositions of the state into pure states with respective entanglement being von neumann entropy of the reduced state . by its
definition calculating @xmath49 is a highly non - trivial optimization problem , which becomes numerically intractable very rapidly if we increase the dimensions of the hilbert spaces .
remarkably , there exist analytical expressions for two - qubit systems as well as for highly symmetric states .
recently , @xmath49 was calculated for symmetric gaussian states of two modes@xcite .
but for general two mode gaussian states , it is still not easy if not impossible to carry out the @xmath49 .
for these reasons the gaussian entanglement of formation ( geof ) @xmath50 is introduced @xcite to quantify the entanglement of bipartite gaussian states by taking the infimum in [ wave1 ] only over decompositions into pure gaussian states .
for any two - mode gaussian state with correlation matrix @xmath11 = @xmath51 , it is proved @xcite that the gaussian entanglement of formation is given by the entanglement of the least entangled pure state with @xmath52=@xmath53 which is such that @xmath54 as a part of the correlation matrix of the entangled pure state , the symmetric @xmath55 matrix @xmath56 can always be written in the form of @xmath57 , \ ] ] then the pure bipartite state with correlation matrix @xmath52 can be constructed by successively applying two local unitary operations to the two mode squeezed vacuum state , where the two local unitary operations are @xmath58 and @xmath59 respectively in simplectic form
. denote @xmath60 , @xmath61 , @xmath62 .
one has @xmath63
_ { p}=\left [ \begin{array}{ll } n & -k \\ -k & m \end{array } \right ] \quad\ ] ] eq .
( [ wave2 ] ) then can be written as @xmath64 & = & 0 , \label{wave3 } \\ \left ( nx+m / x\right ) \cosh r_{g}-2k\sinh r_{g}-[(nm - k^{2})y+1/y ] & = & 0 . \label{wave4}\end{aligned}\ ] ] so that one has @xmath65 clearly @xmath66 is a monotonical increase function of @xmath67 , and when @xmath68 , the minimal value of @xmath66 will be achieved .
one can subsequently obtains @xmath69 by substracting eq.([wave3 ] ) from eq.([wave4 ] ) .
the gaussian entanglement of formation @xmath50 will be @xmath70 where @xmath71 is the bosonic entropy function , and @xmath72 the second term at the right hand side can be written of as @xmath73 with @xmath74 , representing the noise side of the bipartite state . while @xmath75 represents quantum correlation side of the state .
the difference of the two gives the geof squeezing parameter .
the compaison of the geofs for different ratio of average photon numbers @xmath76 is displayed in fig . 1 .
the relative entropy of entanglement for bipartite quantum state @xmath77 is defined by@xcite : @xmath78 where @xmath79 is the set of all disentangled states , and @xmath80 is the relative entropy of @xmath77 with respect to @xmath81 . consider the relative entropy of entanglement of squeezed thermal state @xmath0 , if one chooses a subset of @xmath79 which contain all gaussian separable state , or more specifically all separable squeezed thermal state to substitute the district @xmath79 to carry out the minimum of relative entropy , clearly such minimums are local minimums .
they can be utilized as up bound @xmath82 of global minimum , the relative entropy of entanglement @xmath83 .
denote the district of all separable squeezed thermal states as @xmath84 , then @xmath85 , and @xmath86 , generally speaking the identity can not be achieved . even for relative entropy of pure gaussian state ,
the minimum is achieved by non - gaussian separable state @xcite .
clearly the above way of obtaining up bound by shrinking the district of minimization can be applied to other gaussian state or any other state . in order to obtain the up bound of the relative entropy of entanglement of squeezed thermal state @xmath0 ,
let us first consider the relative entropy of @xmath0 with respect to seperable squeezed thermal state @xmath87 , with @xmath88 .
the von neumann entropy of state @xmath0 is @xcite @xmath89 with @xmath90 @xmath19 .
and one gets @xmath91 by the unitary of @xmath92 .
put @xmath93 in explicit operator form , @xmath94 then the second part of relative entropy will be @xmath95 by utilizing @xmath96 and with the property that operator can be cycled under the trace , after some algebra one obtains @xmath97\log \widetilde{v}_{a } \nonumber \\ & & + [ \frac{v_{b}}{1-v_{b}}\cosh ^{2}\left ( r-\widetilde{r}\right ) + \frac{1}{% 1-v_{a}}\sinh ^{2}\left ( r-\widetilde{r}\right ) ] \log \widetilde{v}_{b}\end{aligned}\ ] ] where @xmath98=0 $ ] is applied .
let us first find out the maximum point of eq .
( [ wave ] ) by partial differentiation with respect to @xmath99 and @xmath100 regardless the fact that @xmath101 , it will be at the point @xmath102 . meanwhile it is noticed that there is no other maximum point . then let us add the condition of @xmath103 , the maximum should be achieved at the edge of @xmath84 , that is @xmath104 .
after one of the parameter is determined , the remain problem is to seek out the maximum with respect to @xmath105 . if the maximum is achieved at @xmath106 , and denote @xmath107 , the up bound of relative entropy of entanglement for squeezed thermal state will be @xmath108\log \widetilde{v}_{a}^{\ast } \nonumber \\ & & + [ \frac{v_{b}}{1-v_{b}}\cosh ^{2}\left ( r-\widetilde{r}^{\ast } \right ) + % \frac{1}{1-v_{a}}\sinh ^{2}\left ( r-\widetilde{r}^{\ast } \right ) ] \log \widetilde{v}_{b}^{\ast } \end{aligned}\ ] ]
one of the most concerned problem in quantum information is the quantum capacities , it is shown that the rate of quantum information transformation is bounded by the maximal attainable rate of coherent information .
if there exists hashing inequality , that is , for any bipartite state the one - way distillable entanglement is no less than coherent information , then one obtains shannon - like formulas for the capacities @xcite .
the coherent information of bipartite state @xmath77 with reductions @xmath109 and @xmath110 is defined as @xcite @xmath111 for @xmath112 and @xmath113 otherwise .
the hypothetical hashing inequality mentioned above is @xmath114 where @xmath115 is forward classical communication aided distillable entanglement of the state .
we know that relative entropy of entanglement @xmath116 is no less than two way distillation of entanglement @xmath117@xcite , the later is no less than one way distillation of entanglement @xmath115 .
combining with hashing inequality one has @xmath118 one the other hand , it is proved that relative entropy is lower bounded by coherent information@xcite , so @xmath116 @xmath119 is always true irrespective of the hypothesis of hashing inequality .
consider the squeezed non - symmetric thermal state @xmath0 , the reduced states @xmath41 and @xmath120 turn out to benot the same .
the entropies of its reduced states are @xmath121 and @xmath122 respectively , where @xmath123 is the average particle number of the reduced state .
this leads to two different coherent information @xmath124 and @xmath125 .
they are @xmath126 the numerical results are shown in the fig .
a subset of quantum gaussian mixed state is given and scrutinizingly investigated .
the so called squeezed non - symmetric thermal state has several merits : it can be described by three parameters , one for quantum correlation , the other two for randomness ; the separability of the state readily follows from comparison of the first parameter and the geometric average of the last two parameters , if the randomness is stronger , the state is separable , otherwise it is entangled ; all the three parameters have range from zero to one .
it is the simplest gaussian mixed state other than the two parameters squeezed thermal state .
the @xmath127 description of state presented in this paper enable the calculation more comparable . with the local unitary operation method ,
i obtain the geof analytically and write it in a simple style with clearly physical meaning . by the nemeric results i find that : as the squeezing parameter tends towards infinitive , the up bound of relative entropy of entanglement tend to coincide with the coherent information , so that one can determine the relative entropy of entanglement , and if the hashing inequality is right , the distillation entanglement can be determined at infinitive squeezing .
|
i study the three parameters bipartite quantum gaussian state called squeezed asymmetric thermal state , calculate gaussian entanglement of formation analytically and the up bound of relative entropy of entanglement , compare them with coherent information of the state . based on the result obtained
, one can determine the relative entropy of entanglement of the state with infinitive squeezing .
|
natural images contain a variety of perceptual information enabling the viewer to infer the three - dimensional shapes of objects and surfaces @xcite .
@xcite observed that surface geometry mainly has three effects on the appearance of texture in images : foreshortening and scaling of texture elements , and a change in their density .
@xcite proposed the slant , the angle between a normal to the surface and a normal to the image plane , as a measure for surface orientation .
stevens amended this by introducing the tilt , the angle between the surface normal s projection onto the image plane and a fixed coordinate axis in the image plane . in this paper , we will directly infer the surface normal from a single image taken under standard perspective projection .
statistical procedures for estimating surface orientation often make strong assumptions on the regularity of texture .
@xcite assumes observed edge directions provide the necessary information , while @xcite consider circular texture elements with uniform intensity .
@xcite consider the bias of the orientation of line elements isotropically oriented on a 3d plane , induced by the plane s orientation under orthographic projection , along with a computational approach related to kanatani s texture moments @xcite .
@xcite locally estimate `` texture distortion '' in terms of an affine transformation of adjacent image patches .
the strong homogeneity assumption underlying this approach has been relaxed by @xcite , to a condition that is difficult to verify in practice .
@xcite eliminates assumptions on the non - local structure of textures ( like homogeneity ) altogether and aims to estimate shape from the deformation of individual texture elements .
@xcite criticize prior work due to the restrictive assumptions related to homogeneity , isotropy , stationarity or orthographic projection , and claim to devise a shape - from - texture approach in the most general form .
their work , however , also relies on estimating the deformation of single texture elements , similar to @xcite .
we propose a general framework for inferring shape from near regular textures , as defined by @xcite , by applying the locally scaled point process model of @xcite .
this framework enables the simultaneous representation of local variability and global regularity in the spatial arrangement of texture elements which are thought of as a marked point process .
we preprocess the image to obtain a probability map representing an unnormalized intensity estimate for the underlying point process , subsequently apply a non - parametric framework to infer the point locations and based on the resulting point pattern , learn the parameters of a locally scaled point process model to obtain a compact description of 3d image attributes .
point process models have previously been applied in image analysis applications where the goal is the detection of texture elements , see e.g. @xcite and references therein .
these approaches usually apply a marked point process framework , with marks describing the texture elements .
such set - ups rely on a good geometric description of individual texture elements , limiting the class of feasible textures . as our goal is not the detection of individual texture elements but the extraction of 3d information
, we omit the modeling of each texture element and infer the latent point locations in a model free manner .
thus , our sole assumption regarding texture element shape is approximate convexity which offers considerable flexibility .
the remainder of the paper is organized as follows .
the next section contains preliminaries on image geometry followed by the method section describing the image preprocessing , the point pattern detection and the point process inference framework .
we then present results for both simulated and real images with near regular textures .
finally , the paper closes with a short discussion section .
let @xmath0 with @xmath1 and @xmath2 , denote a 3d plane with unknown unit normal @xmath3 and distance @xmath4 from the origin .
we assume @xmath3 to be oriented towards the camera , forming obtuse angles @xmath2 with projection rays @xmath5 .
the world coordinates @xmath6 and image coordinates @xmath7 are aligned as shown in fig . [ fig : imaging - setup ] . here
, we denote the image domain by @xmath8 and assume the image to be scaled to have fixed area , @xmath9 .
we consider the basic pinhole camera @xcite and among the internal parameters , we only look at the focal length @xmath10 which depends on the field of view , see fig .
[ fig : imaging - setup ] . as usual , we identify image points and rays of the projective plane through @xmath11 an image point @xmath5 given by meets @xmath12 in @xmath13 with @xmath14 it follows that a point @xmath15 in @xmath12 is related to the image point @xmath5 through @xmath16 is oriented towards the negative @xmath17-halfspace .
the scaled visible image domain is @xmath18 \times [ -1/2,1/2]$ ] .
given the field of view in terms of an angle @xmath19 , we have @xmath20 .
, scaledwidth=40.0% ] a homogeneous texture covering @xmath12 induces an inhomogeneous texture on the two - dimensional image plane with density given by the surface element @xmath21 where @xmath22 denotes the two - dimensional lebesgue measure .
taking , for instance , the fronto - parallel plane @xmath23 results by merely in the constant scale factor @xmath24 , i.e. the homogeneous density @xmath25 .
however , for arbitrary orientation @xmath3 , this factor depends on @xmath5 , as illustrated in fig .
[ fig : regular points ] .
. then quantifies perspective foreshortening and inhomogeneity of the texture , respectively , as observed in the image , and mathematically represents the visually apparent texture gradient .
in a first step , we apply image preprocessing that generates a probability map @xmath26 representing the spatial arrangement of texture elements in the image . to this end , two elementary techniques are locally applied : boundary detection and the corresponding distance transform .
the former step entails either gradient magnitude computation using small - scale derivative - of - gaussian filters @xcite or , for texture elements with less regular appearance , the earth - mover s distance @xcite between local histograms .
inspecting in turn the histogram of the resulting soft - indicator function for boundaries enables one to determine a threshold and apply the distance transform . in our framework ,
the texture elements are regarded as a realization of a marked point process where the underlying point pattern is latent .
the value of the probability map @xmath27 in @xmath28 denotes the probability that one of the latent points is located in @xmath29 . to recover the latent point pattern based on the information in @xmath30 , we first search for local maxima in @xmath30 . that is , for some @xmath31 , let @xmath32 \times [ x_2 - k_1 , x_2 + k_1]$ ] and set @xmath33 we then define a neighbourhood relation on @xmath34 by setting @xmath35 if @xmath36 } y(z ) \ , \geq \ , k_2 \max\{y(x^1 ) , y(x^2)\},\ ] ] where @xmath37 , @xmath38 $ ] denotes the line from @xmath39 to @xmath40 and @xmath41 is a constant with @xmath42
. we may now write @xmath34 as a union of disjoint neighbourhood components , @xmath43 , where each @xmath44 is neighbour with at least one point in @xmath45 . under the assumption that the texture elements are close to convex , two points @xmath39 and @xmath40 in @xmath34 are neighbours if and only if they likely fall within the same texture element . hence
, we estimate the latent point process @xmath46 as @xmath47 formally , a point process can be described as a random counting measure @xmath48 , where @xmath49 is the number of events in @xmath50 for a borel set @xmath50 of the relevant state space , in our context the image domain @xmath8 .
the intensity measure of the point process is given by @xmath51 and the associated intensity function is @xmath52 for a homogeneous point process , it holds that @xmath53 for some @xmath54 , while for an inhomogeneous point process where the inhomogeneity stems from local scaling @xcite we obtain @xmath55 for some scaling function @xmath56 with parameters @xmath57 .
the scaling function @xmath58 acts as a local deformation in that it locally affects distances and areas .
more precisely , @xmath59 , where @xmath60 denotes the @xmath61-dimensional volume measure and @xmath62 its scaled version for @xmath63 .
for identifiability reasons , @xcite propose normalizing @xmath58 to conserve the total area of the state space .
that is , they define the normalizing constant of the scaling function such that @xmath64 @xcite and @xcite specifically consider the exponential scaling function with @xmath65 .
this scaling function is particularly attractive in that locally scaled distances can be calculated explicitely , @xmath66 for any @xmath67 where @xmath68 denotes the euclidean distance and @xmath69 its scaled version .
examples of exponentially scaled distances are given in fig .
[ fig : exponential scaling ] . here
, we employ the density in as a scaling function where we choose spherical coordinates @xmath70 with @xmath71 $ ] and @xmath72 $ ] . the upper limit
@xmath73 restricting the range of the scaling parameter @xmath74 ensures that @xmath2 and therefore depends on the focal length @xmath75 as well as on the size and location of the observation window @xmath8 .
as suggested by @xcite , we normalize the scaling function such that holds .
that is , we solve @xmath76 it follows that @xmath77 a more general result for @xmath78 \times [ b_1 , b_2]$ ] is given in the appendix . under the model in
, the intensity function in becomes @xmath79 with @xmath80 as in . as a byproduct , the unknown plane parameter @xmath4 cancels .
it sets the absolute scale and can not be inferred from a single image .
furthermore , the scaling function is computationally tractable and , as for the exponential scaling discussed above , the scaled distance function is available in closed form , @xmath81 provided that the basic requirement @xmath82 is fulfilled for all @xmath83 .
examples of scaled distances are given in fig.[fig : our scaling ] . when compared with fig .
[ fig : exponential scaling ] , we see that the perspective scaling in results in similar distance scaling as the exponential scaling while it also provides a coherent description of the perspective foreshortening . for a given image , we assume that the focal length @xmath75 is known . it remains to estimate the parameters @xmath84 of the intensity function in based on the estimated point pattern @xmath46 .
the desired 3d image information , the slant and the tilt of the surface , may then be characterized by the scaling parameter estimates @xmath85 and @xmath86 .
the parameter estimation is performed by maximizing the composite likelihood , see e.g. @xcite , that takes the form @xmath87 the maximum composite likelihood estimate for @xmath88 is @xmath89 . for the remaining two parameters the parameters of interest in our setting we maximize the function @xmath90
we first present the results of a simulation study where we analyse sets of 3d point coordinates sampled from either a perfectly regular pattern or a homogeneous poisson processes and subsequently projected onto the 2d - plane @xmath91\times[-1/2 , 1/2]$ ] , see fig .
[ fig : regular points ] and fig .
[ fig : random points ] .
we estimate the scaling parameters associated with the synthetic patterns via the composite likelihood in .
the true parameter values and the corresponding estimates are given in table [ tab : simulation results ] . while the estimation procedure is able to reconstruct the true values with a resonable accuracy , the results are slightly better for the regular patterns than for the random patterns .
these results are representative for several further such examples ( results not shown ) , and we conclude that the composite likelihood is able to identify the scaling parameters of the perspective scaling function irrespective of the second order structure of the point pattern . .true angles and composite likelihood estimates for the surface normals of the simulated point patterns in figures [ fig : regular points ] and [ fig : random points ] .
regular pattern type refers to the images in figure [ fig : regular points ] and poisson type to the images in figure [ fig : random points ] . [ cols="<,^,^",options="header " , ]
this paper introduces a framework for extracting 3d information from a textured 2d image building on the recently developed locally scaled point processes @xcite . the perspective scaling function quantifies perspective foreshortening and the resulting inhomogeneity of the texture .
the framework is quite flexible regarding assumptions on the texture composition in that it only requires the texture elements to be close to convex in shape and it successfully extracts useful information related to camera orientation .
the separation of image preprocessing and point detection on one hand and the estimation procedure for the scaling parameters on the other hand offers great flexibility .
we believe that the locally scaled point process framework can be applied in more general settings to analyse point patterns in images , for instance , as a new additional inference step in the texture detection algorithms discussed in @xcite and references therein . due to the low computational budget of our framework
, it also seems feasible to combine it with image segmentation where 3d information is needed for several segments within an image , each of which might be covered with a different type of texture elements .
there are further considerable avenues for development .
one area for future development is to build a large hierarchical framework where the three inference steps , the image preprocessing , the point detection and the parameter estimation , are joined in an iterative fashion .
a fully bayesian inference framework along the lines of the work of @xcite could also be an alternative to the composite likelihood estimation performed here .
future work will concentrate on embellishing our inference framework .
we thank ute hahn for sharing her expertise . this work has been supported by the german science foundation ( dfg ) , grant rtg 1653 . the work of thordis l. thorarinsdottir and alex lenkoski
was further supported by statistics for innovation , _
sfi_@xmath92 , in oslo . 10
[ 1]`#1 ` blake , a. , marinos , c. ( 1990 ) : shape from texture : estimation , isotropy and moments .
45 , 323380 .
blostein , d. , ahuja , n. ( 1989 ) : shape from texture : integrating texture - element extraction and surface estimation .
ieee trans .
pami-11 , 12331251 .
canny , j. ( 1986 ) : a computational approach to edge detection .
ieee trans .
pami-8 , 679698 .
clerc , m. , mallat , s. ( 2002 ) : the texture gradient equation for recovering shape from texture .
ieee trans .
24(4 ) , 536549 . forsyth , d. ( 2006 ) : shape from texture without boundaries .
int . j. comp .
vision 67(1 ) , 7191 .
gibson , j. ( 1950 ) : the perception of the visual world .
houghton mifflin , boston , ma .
hahn , u. , jensen , e.v .
, van lieshout , m.c . ,
nielsen , l. ( 2003 ) : inhomogenous spatial point processes by location - dependent scaling . adv .
( sgsa ) 35 , 319336 .
hartley , r. , zisserman , a. ( 2000 ) : multiple kanatani , k. ( 1989 ) : shape from texture : general principle .
38 , 148 .
lafarge , f. , gimelfarb , g. , descombes , x. ( 2010 ) : geometric feature extraction by a multi - marked point process .
ieee trans .
32(9 ) , 15971609 .
liu , y. , hel - or , h. , kaplan , c. , van gool , l. ( 2009 ) : computational symmetry in computer vision and computer graphics . found .
trends comp .
graphics and vision 5(1 - 2 ) , 1195 .
loh , a. , hartley , r. ( 2005 ) : shape from non - homogeneous , non - stationary , anisotropic , perspective texture . in : proc .
. 6978 .
malik , j. , rosenholtz , r. ( 1997 ) : computing local surface orientation and shape from texture for curved surfaces .
int . j. comp .
vision 23(2 ) , 149168 .
mller , j. ( 2010 ) : spatial point patterns : parametric methods . in : gelfand , a.e . ,
diggle , p.j . , fuentes , m. , guttorp , p. ( eds . )
handbook of spatial statistics .
crc press , boca raton , fl .
mller , j. , waagepetersen , r.p .
( 2004 ) : statistical inference and simulation for spatial point processes .
chapman & hall / crc , boca raton , fl .
pele , o. , werman , w. ( 2009 ) : fast and robust earth mover s distances . in : proc .
conf . comp .
vision ( iccv ) .
prokeov , m. , hahn , u. , jensen , e.b.v .
( 2006 ) : statistics for locally scaled point processes . in : baddeley , a. , gregori , p. , mateu , j. , stoica , r. , stoyan , d. ( eds . ) case studies in spatial point process modelling .
springer , new york .
rajala , t. , penttinen , a. ( 2012 ) : bayesian analysis of a gibbs hard - core point pattern model with varying repulsion range . comp .
data anal . in press .
stevens , k.a .
( 1980 ) : surface perception from local analysis of texture and contour .
tech . rep .
ai - tr 512 , mit technical report , artificial intelligence laboratory .
tuceryan , m. , jain , a.k .
( 1998 ) : texture analysis . in : chen ,
pau , l.f . ,
wang , p.s.p .
handbook of pattern recognition and computer vision ( 2nd edition ) .
207248 . world scientific , singapore .
witkin , a.p .
( 1981 ) : recovering surface shape and orientation from texture .
17 , 1745 .
in our data analysis , we assume that the image domain is normalized such that @xmath18 \times [ -1/2,1/2]$ ] . more generally , the image domain could be of the form @xmath93 \times [ b_1,b_2]$ ] for some @xmath94 with @xmath95 and @xmath96 . in this case
, the condition of conservation of the total area in becomes @xmath97 it follows that @xmath98
|
shape from texture refers to the extraction of 3d information from 2d images with irregular texture .
this paper introduces a statistical framework to learn shape from texture where convex texture elements in a 2d image are represented through a point process . in a first step ,
the 2d image is preprocessed to generate a probability map corresponding to an estimate of the unnormalized intensity of the latent point process underlying the texture elements .
the latent point process is subsequently inferred from the probability map in a non - parametric , model free manner .
finally , the 3d information is extracted from the point pattern by applying a locally scaled point process model where the local scaling function represents the deformation caused by the projection of a 3d surface onto a 2d image .
_ keywords : _
3d scenes , convex texture elements , locally scaled point processes , near regular texture , perspective scaling , shape analysis
|
when considering physics beyond the standard model ( sm ) , supersymmetry , in the form of the minimal supersymmetric standard model ( mssm ) is usually considered as one of the most appealing options . however , due to its vast parameter space , a standard procedure is to assume universality conditions of the soft terms at the gut scale , and allow the renormalization group equations ( rges ) to break the universality when running down to the weak scale .
this constrained realization is named cmssm , and leads to flavoured soft terms acquiring a structure following minimal flavour violation ( mfv ) @xcite .
mfv effectively suppresses all susy contributions to flavoured processes , solving the so - called susy flavour problem .
however , it also leads to sfermion masses with small splittings .
this can cause issues in the solution of the hierarchy problem , which depends mainly on having light third generation squark masses .
the issue is caused due to the lack of susy signals at the lhc , which forces the masses of the first two generation squarks to be pushed to larger values .
the mfv structure causes the first two generations to drag with them the mass of the third generation squarks to these larger values , compromising in this way the solution of the hierarchy problem .
one would therefore like to have a framework similar to mfv , such that contributions of flavoured processes are still suppressed to some degree , but allowing the third generation masses to be decoupled from the first two .
one would also hope that the flavour suppression in this new framework be somewhat milder , such that some significant effects could be observed in flavour experiments .
in particular , one would hope these new effects could solve the ongoing tension between cp violating @xmath1 observables in the @xmath2 and @xmath3 sectors @xcite .
a framework based on a @xmath0 symmetry in the quark sector was built in @xcite , and was found to satisfy all of the above requirements .
further studies of this symmetry were carried out in @xcite , where the squark and slepton phenomenology was analysed in further detail , both in and out of susy .
nevertheless , in these works the symmetry structure was applied directly at the electroweak scale , with no consideration of rge running .
as one would expect that the breaking of such a symmetry would occur at a high scale , it was found necessary to study the consequences of the running on the low - scale structures . in particular , it was unclear if the virtues of @xmath0 would be maintained at the weak scale , and what kind of initial conditions could lead to the required split scenario .
the current paper is a summary of the work carried out in @xcite . in section [ sec : rev ]
we briefly review the @xmath0 framework and the conditions leading to the solution of the flavour tension . in section [ sec : rge ] we show the result of our study of the rge evolution , and then conclude .
we shall restrict ourselves to the quark sector .
in contrast to mfv , which assigns a @xmath4 symmetry for each flavoured chiral supermultiplet , this framework assigns a @xmath5 symmetry for each supermultiplet , under which only the first two generations shall transform . in the limit of exact flavour symmetry ,
the only terms allowed in the superpotential involving flavoured fields are those of the third generation ( s)quarks : @xmath6 as in mfv , we proceed to break the flavour symmetries by adding spurion fields .
the minimal way of constructing the yukawas is by adding bi - doublet spurions @xmath7 and @xmath8 , from which we build the structure of the @xmath9 block , and a doublet @xmath10 , transforming only under @xmath11 , used to connect the first two generations with the third . with these assumptions , the yukawas
acquire the structure : @xmath12 with everything above the horizontal dashed line having two rows , and everything to the left of the vertical dashed line having two columns .
we find it possible to parametrize the whole matrices in terms of the eigenvalues of the bi - doublets , @xmath13 and @xmath14 , a left - mixing matrix with one angle and phase per bi - doublet , @xmath15 and @xmath16 , a suppression parameter @xmath17 for the doublet , and two complex couplings @xmath18 .
notice only three phases are independent . with this setup
, one can build the ckm matrix , and fit the following parameters @xcite : @xmath19 where @xmath20 .
we choose to fix @xmath21 . after determining which spurions are needed to build the yukawa matrices
, we follow the mfv assumption demanding all flavour structures to depend exclusively on these spurions . with this , we find that the soft terms acquire the following form : @xmath22 where @xmath23 , and all @xmath24 and @xmath25 parameters are real , of @xmath26 . these structures lead to the following contributions to @xmath1 observables : @xmath27 where : @xmath28~ , \qquad x_g = \frac { m^2_{\tilde g } } { m_{q_3}^2 } ~ , \label{eq : f0 } \\
f_{0}(x_g ) & = & \frac{11 + 8 x_g -19x_g^2 + 26 x_g\log(x_g)+4x_g^2\log(x_g)}{3(1-x_g)^3}~ , \qquad x=\frac{s_l^2 c^2_d}{|v_{ts}|^2}\end{aligned}\ ] ] with @xmath29 being the typical one - loop function of the sm to @xmath1 processes .
the @xmath30 parameters are defined in @xcite , and are related to the @xmath17 and @xmath31 terms . as a function of @xmath32 and @xmath33 .
white region has @xmath34 , grey region has @xmath35 .
red and blue lines are rough limits on @xmath32 , taken from atlas , green line is a rough limit on @xmath33 , taken from cms.,scaledwidth=40.0% ] with these modifications , one can find that the @xmath0 framework can provide a large enough contribution to @xmath36 and @xmath37 in the correct directions , such that the flavour tension mentioned above is solved . for not too large @xmath38 , we required @xmath39 , but found that values of @xmath40 as low as @xmath41 could also reduce the tension by compensating with slightly large @xmath26 parameters .
this implies upper bounds on gluino and sbottom masses , as shown in figure [ fig : f0 ] .
figure [ fig : f0 ] also shows the current status of the @xmath0 framework in susy .
the red and blue lines indicate lower bounds on @xmath32 set by the atlas collaboration , for the cases @xmath42 @xcite and @xmath43 @xcite , respectively .
the green line indicates lower bounds on @xmath33 by the cms collaboration @xcite .
the white area shows @xmath34 , and is almost completely excluded by direct searches .
the grey area has @xmath44 , and is not heavily constrained when @xmath43 .
thus , current data favours heavy gluinos , and somewhat large mixing parameters .
we now turn to study the properties of this framework when defined at a high energy scale , and then evolved down to the electroweak scale .
we address three questions : * ( 1 ) * what sort of initial conditions should we look for ? * ( 2 ) * how does the mixing evolve ? * ( 3 ) * are the virtues of @xmath0 preserved at all scales ? to simplify the study of the initial conditions , we shall turn to a cmssm - like setup . at the gut scale ,
we keep the universal gaugino mass , @xmath45 , but differentiate between the common higgs mass , @xmath46 , the common ( heavy ) first two generations scalar mass , @xmath47 , and the common ( light ) third generation scalar mass , @xmath48 .
moreover , we use the parameter @xmath49 instead of @xmath48 .
the a - terms follow a structure similar to that of the yukawas , with different @xmath26s and multiplied by a factor @xmath50 . after running the parameters down to the electroweak scale , we require no tachyonic states , radiative electroweak symmetry breaking , and the correct higgs mass within @xmath51 .
we also prefer regions where @xmath35 , and attempt to reduce the hierarchy problem by having small values of @xmath52 and @xmath53 . .
the dark regions correspond to no ewsb , while the orange regions have a tachyonic stop .
the green , dashed lines delimitate the regions within 1 sigma of the higgs mass , and the red , solid curve indicates the regions below which the flavour tension could be solved .
the blue dot represents the benchmark.,title="fig:",scaledwidth=32.0% ] .
the dark regions correspond to no ewsb , while the orange regions have a tachyonic stop .
the green , dashed lines delimitate the regions within 1 sigma of the higgs mass , and the red , solid curve indicates the regions below which the flavour tension could be solved .
the blue dot represents the benchmark.,title="fig:",scaledwidth=32.0% ] .
the dark regions correspond to no ewsb , while the orange regions have a tachyonic stop .
the green , dashed lines delimitate the regions within 1 sigma of the higgs mass , and the red , solid curve indicates the regions below which the flavour tension could be solved .
the blue dot represents the benchmark.,title="fig:",scaledwidth=32.0% ] + although in @xcite we studied two different benchmarks , here we shall concentrate only in one , and describe the parameter space around this point .
the benchmark is characterized by @xmath54 tev , @xmath55 tev , @xmath56 tev , @xmath57 and @xmath58 .
we show in figure [ fig : benchmark1 ] the parameter space around this point , with the different constraints represented in different colours .
we can see that the region that fulfills all constraints is very small , especially if we demand @xmath59 tev .
the benchmark leads to values of @xmath52 around 500 gev , and very specific splittings . in the stop sector
, we find @xmath60 , which leads to an average stop mass of about @xmath61 tev .
nevertheless , as the stop mixing is large , we find the mass of the lightest stop to be lower than 500 gev . on the other hand , in the sbottom sector
, we have @xmath62 , leading to an average sbottom mass of @xmath63 tev .
notice that this setup involves a very mild splitting at the gut scale , but can lead to a larger splitting in the stop sector .
having found a point with suitable initial conditions , we now turn to the study of the mixing . to this end , we use objects directly related to physical observables , which are also connected with the mixing , and study their rge evolution . as our interest lies on @xmath1 processes , which receive their main contribution from @xmath64 operators , the objects in question are defined as @xmath65 where @xmath66 is the diagonalization matrix of @xmath67 in the basis of diagonal down quarks . in @xcite
, it was shown that the @xmath0 structure leads to @xmath68 , @xmath69 and @xmath70 to be closely related : @xmath71 where @xmath72 .
thus , assuming for now that these relations are held throughout the running , we shall only need to study the evolution of one of these objects , which we choose to be @xmath70 .
( left ) and arg@xmath73 ( right ) . on the left
, we show @xmath74 in blue , green and red . in every region
the lighter colour correspond to @xmath75 fixed to @xmath76 . on the right we fix @xmath77 , with @xmath78 in blue , red , green , grey and magenta , respectively .
the lighter regions correspond to @xmath79 . in the first two plots the dashed brown lines mark the region where
the flavour tension can be solved .
on the left , the region is above the line , while and on the right it is between the two lines.,title="fig:",scaledwidth=45.0% ] ( left ) and arg@xmath73 ( right ) . on the left
, we show @xmath74 in blue , green and red . in every region
the lighter colour correspond to @xmath75 fixed to @xmath76 . on the right we fix @xmath77 , with @xmath78 in blue , red , green , grey and magenta , respectively .
the lighter regions correspond to @xmath79 . in the first two plots the dashed brown lines mark the region where the flavour tension can be solved . on the left ,
the region is above the line , while and on the right it is between the two lines.,title="fig:",scaledwidth=45.0% ] in figure [ fig : lambdasb1 ] we show how @xmath70 evolves with the scale .
the main result is that , although there exists some suppression effect in the magnitude , the mixing remains relatively stable throughout the running .
the figure also shows the minimum value that the mixing must have in order to solve the flavour tension , according to @xcite . here , we can see that the @xmath26 parameters can not be too small , which is a direct consequence of the small value of @xmath40 .
finally , we need to check if the @xmath0 relations of equation ( [ eq : lambda ] ) are kept after the rge running . the best way to do this is to use the @xmath80 to define new objects that should be approximately rge invariant , at least within their theoretical errors , and check if this invariance holds .
we choose : @xmath81 the first ratio tests the correlations between the @xmath3 and @xmath82 sectors , the second ratio tests those between the @xmath2 and @xmath3 sectors .
if we find these ratios to hold within their theoretical errors , we can consider the @xmath0 symmetry to be preserved by the running
. however , we first need to write the theoretical error for each of these quantities . the main source of error in the absolute value of each ratio lies on a nlo effect due to @xmath83 , which can generate deviations of up to @xmath84 the value of the ratio . on the other hand , the phase has a fixed correction , of order @xmath85 .
we shall then take the error on the phase equal to @xmath86 , after including the correction .
moreover , we neglect the errors due to the running of the ckm elements in @xmath87 . .
the dashed lines correspond to our estimated theoretical uncertainty .
we show results for @xmath74 in brown , magenta and blue , respectively.,title="fig:",scaledwidth=45.0% ] .
the dashed lines correspond to our estimated theoretical uncertainty .
we show results for @xmath74 in brown , magenta and blue , respectively.,title="fig:",scaledwidth=45.0% ] the result of evolving these ratios down to the electroweak scale is shown in figure [ fig : invariant ] .
different initial values for @xmath88 are shown in different colours .
the main conclusion is that in most cases the ratios are preserved within their theoretical errors . in general ,
the largest the initial value of @xmath88 , the better the ratios are preserved .
given the fact that the @xmath0 relations are maintained at all scales , the stability when running @xmath70 implies that both @xmath68 and @xmath69 also have a stable behaviour when solving the rges .
this means that the virtues of the framework are preserved at all scales .
there exists a tension between the solution of the hierarchy problem , the current lhc bounds on sparticle masses , and the use of mfv .
furthermore , there is a slight tension between cpv observables in the @xmath2 and @xmath3 sectors , which mfv can not solve .
this encourages the use of other frameworks with suppression properties similar to those of mfv , but with the capacity of solving all of these tensions . in this work ,
we have reviewed a @xmath0 framework in susy , originally proposed in @xcite .
we wrote the main ingredients of this framework , and explored the regions in the @xmath89 parameter space where the flavour tension would be solved .
we showed that current direct search bounds constrain significantly this parameter space , forcing us to consider regions which require slightly large @xmath26 parameters .
we also presented several features of this framework when defined at a high scale @xcite .
we considered the definition of the initial conditions , the evolution of the mixing , and the preservation of relations between contributions to observables in the @xmath2 , @xmath3 and @xmath82 sectors .
we defined a benchmark point and studied the parameter space around it , finding it strongly constrained .
however , we also found that the mixing was rather stable during the rge evolution , and that the @xmath0 relations between different observables were preserved throughout the running .
the author would like to thank the organizers of the discrete 2012 conference .
the author also acknowledges support from the grants generalitat valenciana vali+d , spanish mineco fpa 2011 - 23596 and the generalitat valenciana prometeo - 2008/004 , and would like to thank gianluca blankenburg , who collaborated in the main topic of this talk .
99 g. dambrosio , g. f. giudice , g. isidori and a. strumia , nucl .
b * 645 * ( 2002 ) 155 [ hep - ph/0207036 ] .
w. altmannshofer , a. j. buras , s. gori , p. paradisi and d. m. straub , nucl .
b * 830 * ( 2010 ) 17 [ arxiv:0909.1333 [ hep - ph ] ] .
r. barbieri , g. isidori , j. jones - perez , p. lodone and d. m. straub , eur .
j. c * 71 * ( 2011 ) 1725 [ arxiv:1105.2296 [ hep - ph ] ] .
r. barbieri , p. campli , g. isidori , f. sala and d. m. straub , eur .
j. c * 71 * ( 2011 ) 1812 [ arxiv:1108.5125 [ hep - ph ] ] ; r. barbieri , d. buttazzo , f. sala and d. m. straub , jhep * 1207 * ( 2012 ) 181 [ arxiv:1203.4218 [ hep - ph ] ] ; a. j. buras and j. girrbach , jhep * 1301 * ( 2013 ) 007 [ arxiv:1206.3878 [ hep - ph ] ] ; g. blankenburg , g. isidori and j. jones - perez , eur .
j. c * 72 * ( 2012 ) 2126 [ arxiv:1204.0688 [ hep - ph ] ] ; r. barbieri , d. buttazzo , f. sala and d. m. straub , jhep * 1210 * ( 2012 ) 040 [ arxiv:1206.1327 [ hep - ph ] ] .
g. blankenburg and j. jones - perez , eur .
j. c * 73 * ( 2013 ) 2289 [ arxiv:1210.4561 [ hep - ph ] ] .
[ atlas collaboration ] , atlas - conf-2012 - 107 .
g. aad _ et al . _ [ atlas collaboration ] , eur .
j. c * 72 * ( 2012 ) 2174 [ arxiv:1207.4686 [ hep - ex ] ] .
s. chatrchyan _ et al . _
[ cms collaboration ] , jhep * 1301 * ( 2013 ) 077 [ arxiv:1210.8115 [ hep - ex ] ] .
|
from the naturalness point of view , the first lhc results seem to disfavour any constrained mssm realization with universal conditions at the susy - breaking scale . a more motivated scenario is given by split - family susy ,
in which the first two generations of squarks are heavy , compatible with a @xmath0 flavour symmetry . here , after reviewing the flavour structures obtained in this framework , we consider the flavour symmetry to be broken at a very high scale , and study the consequences at low energies through its rge evolution .
initial conditions compatible with a split scenario are found , and the preservation of correlations from minimal @xmath0 breaking are checked .
|
high - redshift damped lyman-@xmath5 systems ( dlas ) are the highest column density class of qso absorption lines .
defined by @xmath10 ( wolfe et al .
1986 ) , these systems are thought to trace the interstellar medium ( ism ) of high - redshift galaxies . dedicated surveys over the past two decades
have helped trace the global properties of high - redshift dlas , including their contribution to the cosmological baryon density ( storrie - lombardi & wolfe 2000 ; prochaska & herbert - fort 2004 ) , their chemical enrichment ( e.g. , prochaska et al .
2003 ) , their dust content ( e.g. , pettini et al .
1994 ) , and molecular fraction ( ledoux , petitjean , & srianand 2003 ) .
these studies have demonstrated that the dlas have a baryonic mass density comparable to the mass density of modern galaxy disks , that the metallicity of high-@xmath11 dlas is slowly increasing , and that the majority of dla sight lines have low dust - to - gas ratios and molecular fractions .
understanding the detailed physics of the ism in dlas is an important step in understanding high - redshift galaxies in general .
wolfe , prochaska , & gawiser ( 2003a ) have constructed detailed models for the thermal equilibrium of the ism in a set of dlas , calculating the heating rate experienced by the gas due to the ultraviolet emission from young hot stars .
they used observations of absorption out of the @xmath2 level of c@xmath1ii ( hereafter c@xmath1ii@xmath6 ) a direct indicator of the cooling rate through [ c@xmath1ii ] 158 @xmath12 m emission ( pottasch , wesselius , van duinen 1979 ) to infer the actual heating rate experienced by the gas ( assuming thermal equilibrium ) .
their comparison of the inferred and calculated heating rates suggests dlas harbor significant star formation .
the models of wolfe et al .
( 2003a , b ) required that the observed c@xmath1ii@xmath6 in dlas arise in a cold neutral medium ( cnm ) , i.e. , in gas with temperatures @xmath13 k ; their wnm models give sfrs that violate observations of the bolometric background .
the detection of @xmath14 absorption in some high-@xmath11 dlas is further evidence that at least some of these systems contain a cnm ( ledoux et al .
2003 ; hirashita & ferrara 2005 ) .
however , such temperatures are at odds with 21-cm absorption studies ; all @xmath15 dlas searched for 21-cm absorption show @xmath16k ( @xmath7 ) ( kanekar & chengalur 2003 ) .
wolfe et al . ( 2003b ) argue that this discrepancy is likely due to the differing properties of the sight lines probed by the optical background sources and by the more extended radio sources . in this letter
we present a method for determining the kinetic temperature of interstellar matter based solely on basic atomic physics .
our method compares the excitation of the upper @xmath2 fine - structure levels in si@xmath1ii and c@xmath1ii , which have excitation energies that differ by a factor of four .
we describe our approach in [ sec : method ] .
we apply this technique to the @xmath17 dla toward the quasar pss 1443 + 27 in [ sec : psscnm ] , demonstrating that this dla contains a substantial reservoir of cold gas .
lastly , we discuss the implications of this temperature determination in [ sec : discussion ] .
in this section we discuss how analysis of the @xmath0 fine structure excitation in si@xmath1ii and c@xmath1ii can be used to limit interstellar kinetic temperatures .
srianand & petitjean ( 2000 ) have used a similar analysis to limit temperatures and densities in an associated absorber .
silva & viegas ( 2002 , hereafter sv ) give a detailed summary of the excitation equilibrium of fine structure lines in dlas .
we treat the excitation of the @xmath0 fine structure states as a two - level atom .
the equilibrium ratio of the densities in the upper and lower levels ( spitzer 1978 ) , @xmath20 and @xmath21 , respectively , is @xmath22 the quantities @xmath23 , @xmath24 , and @xmath25 are the familiar einstein transition probabilities .
the energy density of the cosmic microwave background , @xmath26 , for direct excitation of the transitions is calculated assuming a standard cosmology , i.e. , @xmath27 , where @xmath28 k ( mather et al . 1999 ) .
the fluorescent rates @xmath29 due to the great opacity of the ground - state transitions of interest ( sarazin , rybicki , & flannery 1979 ; see wolfe et al .
the summations describe excitations and deexcitations with collision partners @xmath30(@xmath31 , @xmath32 , and h@xmath33 ) , where @xmath34 is the particle density of each partner and @xmath35 and @xmath36 are the maxwellian - averaged collision rate coefficients for excitation and deexcitation . the collision rates are related to one another and to @xmath37 , the collision strength ( spitzer 1978 ) , by : @xmath38 where @xmath39 is the energy of the @xmath2 level above the ground state , and @xmath40 is the gas kinetic temperature .
we adopt atomic data from the same sources as sv throughout . the column density of material in the @xmath2 level of c@xmath1ii is @xmath41 where @xmath42 is the gas - phase abundance of carbon , @xmath43 is the density of neutral hydrogen , @xmath44 is the differential pathlength along the line of sight , and the ratio @xmath45 is a function of four quantities : @xmath46 , @xmath43 , @xmath47 , and @xmath40 ( eqn . [ eqn : populations ] ) .
we have assumed @xmath48 , consistent with limits on in dlas ( see figure 12 of wolfe et al .
a similar expression applies for si@xmath1ii@xmath6 .
the column densities of c@xmath1ii@xmath6 and si@xmath1ii@xmath6 are therefore density- and excitation - weighted integrals over pathlength . to understand the usefulness of si@xmath1ii@xmath6 and c@xmath1ii@xmath6 to limit gas temperatures , imagine that the excitation of the @xmath2 states were only due to electron collisions and deexcitation due to spontaneous emission .
in this case @xmath49 \times [ \omega_{12}^e(t)_{\rm si}/\omega_{12}^e(t)_{\rm c } ] \times \exp\{-(t_{12}^{\rm si } - t_{12}^{\rm c})/t\}$ ] , i.e. , the ratio is only a function of temperature and the si / c ratio .
figure [ fig : ratio ] shows the ratio @xmath50 as a function of temperature , including all of the terms in eqn .
( [ eqn : populations ] ) , for several densities at @xmath51 assuming [ si / c]=0.@xmath52 . ]
@xmath50 varies slowly with temperature for @xmath53 k ( for all but the highest densities ) and more strongly at lower temperatures .
the sensitivity to density is modest , especially for low @xmath54 , because the populations of the @xmath2 levels of si@xmath1ii and c@xmath1ii depend on density in a similar manner .
we use the ratio of the @xmath2 to @xmath55 levels of c@xmath1ii and si@xmath1ii to constrain the allowable range of densities .
gas at @xmath56 k produces ratios @xmath57 ; the corresponding optical depth ratio of the 1264.738 line of si@xmath1ii@xmath6 to the typically - unresolved 1335.663 and 1335.708 doublet of c@xmath1ii@xmath6 is @xmath58 ( @xmath59-values from morton 2003 ) .
thus , the use of this ratio to distinguish cool from warm temperatures requires large c@xmath1ii@xmath6 optical depths and good s / n at si@xmath1ii@xmath6 .
observations of @xmath50 will allow a range of parameters @xmath60 , and @xmath40 , which apply to the c@xmath1ii@xmath6-bearing gas .
however , the maximum allowable temperature can be low if the limits on @xmath50 are low .
prochaska et al . ( 2001 ) present high - resolution ( @xmath61 ) , high s / n observations of the @xmath4 dla toward of the quasar pss 1443 + 27 using hires ( vogt et al .
1994 ) on the keck i 10-m telescope .
the properties of this dla are summarized in table [ tab : properties ] . due to its redshift and relatively strong metal lines , column densities
can only be measured for a few species .
table [ tab : properties ] gives a measurement of fe@xmath1ii ( from the 1611.2 transition ) and meaningful limits on c@xmath1ii@xmath6 , si@xmath1ii and si@xmath1ii@xmath6 .
all but the last of these are from prochaska et al .
( 2001 ) .
figure [ fig : continuum ] shows the hires spectrum at the expected location of si@xmath1ii@xmath6 in this absorber , including our fitted continuum .
our limit to si@xmath1ii@xmath6 was determined using an empirically - measured signal - to - noise ratio ( following sembach & savage 1992 ) and includes continuum placement uncertainties .
figure [ fig : profile ] shows the normalized profiles of si@xmath1ii@xmath6 and c@xmath1ii@xmath6 . also shown
are models for the expected si@xmath1ii@xmath6 absorption for a canonical warm neutral medium ( wnm ; with @xmath62 @xmath63 , @xmath64 , and @xmath65 k ) and cnm ( with @xmath66 @xmath63 , @xmath67 , and @xmath68 k ) .
the models are derived by scaling the c@xmath1ii@xmath6 profile by the amount predicted using eqns .
( [ eqn : populations ] ) and ( [ eqn : cstarcolumn ] ) and assuming [ si / c]@xmath69 .
the hypothesis that c@xmath1ii@xmath6 absorption toward pss 1443 + 27 arises in a canonical wnm is not consistent with the observations . to limit @xmath40
, we calculate the excitation balance of the fine structure levels of si@xmath1ii and c@xmath1ii for densities @xmath70 and @xmath71 and temperatures @xmath72 .
we compare these calculations with our observational limits for @xmath50 , @xmath73 , and @xmath74 to constrain the physical properties ( @xmath43 , @xmath54 , and @xmath40 ) of the gas in the c@xmath1ii@xmath6-bearing gas . and
@xmath75 values are allowable , and only t is stringently constrained in our analysis . ]
the allowable range of physical conditions in the pss 1443 + 27 dla then depends upon the ratios of si / c and c / fe , which are used for determining the expected @xmath50 and @xmath76 , respectively .
c@xmath1ii is rarely measurable in dlas due to saturation effects ; we estimate @xmath77 from @xmath78 ( table [ tab : properties ] ) with an assumed abundance ratio . as a fiducial
, we adopt the solar system ratios from grevesse & sauval ( 1998 ) : @xmath79 and @xmath80 .
this gives @xmath81 .
nucleosynthetic effects are unlikely to modify the c / fe ratio , as a roughly solar value is found in low - metallicity milky way stars ( e.g. , carretta , gratton , & sneden 2000 )
. however , nucleosynthetic effects could enhance @xmath5 element abundances , leading to high si / c ratios ( see , e.g. , prochaska 2004 ) .
in one case where the c@xmath1ii measurements may be reliable , dodorico & molaro ( 2004 ) find @xmath82 and @xmath83 ( assuming the fe@xmath1ii column from lu et al .
this implies [ c / fe]@xmath84 and [ si / c]@xmath85 compared with our adopted solar system abundances .
the overall metallicity of this absorber is [ si / h]@xmath86 , however , which is significantly lower than the metallicity of the dla studied in this work .
the presence of dust will raise the c / fe ratio due to differential depletion .
the depletion of fe in dlas is typically modest , similar to the `` halo '' and `` disk - halo '' clouds in the milky way ( savage & sembach 1996 ) with [ fe / h]@xmath87.@xmath88 .
thus , unlike the case in dlas , the sub - solar gas - phase fe / h quoted here is due entirely to depletion into dust grains .
] for diffuse clouds in the milky way , sofia et al .
( 2004 ) find [ c / h]@xmath89 .
thus , we expect [ c / fe]@xmath90 . in low - depletion milky way clouds , and by extension dlas ,
si is likely not depleted much differently from c , and we assume [ si / c]@xmath91 .
the small number of species for which measurements are possible in the dla toward pss 1443 + 27 make it difficult to judge the degree of fe depletion .
moderate depletion is suggested by [ si / fe]@xmath92 , but this could also result from @xmath5-enhanced abundances .
we give temperature limits for the c@xmath1ii@xmath6-bearing gas in the @xmath93 dla toward pss 1443 + 27 for six different abundance assumptions in table [ tab : results ] .
the first three assume [ si / c]@xmath69 with [ c / fe]@xmath94 , + 0.3 , and + 0.6 , allowing for modest levels of depletion .
the next three assume [ si / c]@xmath95 , i.e. , @xmath5-enhanced abundances ( see prochaska 2004 ) .
we note that the model in which [ si / c]@xmath96[c / fe]@xmath97 is not self - consistent : scaling both si@xmath1ii and fe@xmath1ii by solar abundances gives inconsistent values for the c@xmath1ii column density .
the highest limit is @xmath8 k , assuming assuming [ c / fe]@xmath98 and [ si / c]@xmath99 .
this is likely an extreme assumption for the c / fe ratio .
the highest temperature limit for @xmath5-enhanced abundances , which we prefer due to the distribution of [ si / fe ] at very low metallicities ( where dust depletion effects are likely small ; see prochaska 2004 ) , is @xmath9 k. temperatures above 1400 k , consistent with @xmath100 limits for the @xmath101 dlas observed at 21-cm ( kanekar & chengalur 2003 ) , require [ c / fe]@xmath102 to + 0.8 for both solar and @xmath5-enhanced abundances .
we feel this is inconsistent with our knowledge of differential depletion in the milky way and dlas .
we have presented a method for limiting gas temperatures in the ism of galaxies through measurements and analysis of the @xmath2 fine structure levels of si@xmath1ii and c@xmath1ii .
we have applied this method to limit the properties of the c@xmath1ii@xmath6-bearing gas in the @xmath4 dla toward pss 1443 + 27 .
we rule out the hypothesis that the c@xmath1ii@xmath6 absorption arises in a wnm .
our conservative temperature limit for this gas is @xmath103k ; we obtain stricter temperature limits if [ si / c]@xmath104 or [ c / fe]@xmath105 : @xmath106 k. the detection of a cnm ( wolfire et al .
1995 ) in a high - redshift dla is significant : while the gas seen in this dla is likely not associated with the dense star - forming clouds , our result demonstrates that the physical conditions of the ism in this system do not preclude the existence of cold material , including , in principle , the dense clouds from which stars could form .
the detection of a cnm in the z=4.224 absorber toward pss 1443 + 27 is the first detection at such a large redshift .
we stress that our measurements allow a wnm as part of a multiphase medium toward pss 1443 + 27 , but the majority of the c@xmath1ii@xmath6 can not come from warm material .
the existence of cnm material may be a feature of many high-@xmath11 dlas , as suggested by wolfe et al .
( 2003a , b ) .
while @xmath107 h@xmath1i 21-cm absorption - line measurements suggest the presence of cold h@xmath1i in dlas , no 21-cm absorption has been found in dlas with @xmath15 ( kanekar & chengalur ) .
dlas have been observed ; pss 1443 + 27 has not been searched for 21-cm absorption . ]
we note , however , that at least three @xmath108 dlas show @xmath14 absorption ( ledoux et al .
2003 ) , indicating the presence of cold gas ( hirashita & ferrara 2005 ) . searching for @xmath14 toward pss 1443 + 27 might be difficult due to the strength of the lyman-@xmath5 forest at @xmath109 . in the future
we will apply our method for determining gas kinetic temperatures to dlas for which 21-cm and @xmath14 measurements have been attempted .
we note that the dla toward pss 1443 + 27 may not be typical .
it shows a high metallicity ( [ si / h]@xmath110 ) , especially compared with other @xmath111 dlas ( prochaska et al .
furthermore , we chose this dla for this experiment because it has very strong c@xmath1ii@xmath6 .
it has the highest c@xmath1ii@xmath6/h@xmath1i ratio measured , some @xmath112 dex above the next highest ( the @xmath113 dla toward q2206 - 19 ; wolfe et al .
the intensity of radiation in this dla calculated following wolfe et al .
( 2004 ) is higher than all other systems .
although the fuel for star formation may be present in this dla , no optical counterpart to this dla has been identified in deep ground - based and _ hubble space telescope _ images of pss 1443 + 27 ( prochaska et al .
2002 ; l. storrie - lombardi , private communication ) .
the method presented here will be discussed further , and temperature limits given for a larger number of dlas , in howk et al .
we note that the use of the relative populations of the @xmath2 levels of si@xmath1ii and c@xmath1ii may also be useful for constraining temperatures in milky way gas .
the authors wish to recognize and acknowledge the very significant cultural role and reverence that the summit of mauna kea has always had within the indigenous hawaiian community .
we are most fortunate to have the opportunity to conduct observations from this mountain .
we thank an anonymous referee for comments that improved this work .
jch and amw recognize support from proposal number hst - ar-09931.01-a , provided by nasa through a grant from the space telescope science institute , which is operated by the association of universities for research in astronomy , incorporated , under nasa contract nas5 - 26555 .
amw and jxp are supported by nsf grant ast 03 - 07824 .
lr @xmath46 & 4.2240 + @xmath114 & @xmath115 + @xmath116 $ ] & @xmath117 + @xmath118 & @xmath119 + @xmath120 & @xmath121 + @xmath122 & @xmath123 + @xmath124 & @xmath125 + @xmath126 & @xmath127 + @xmath128 & @xmath129 + ccccc 1 & + 0.0 & + 0.0 & @xmath130 & 1.44 + 2 & + 0.0 & + 0.3 & @xmath131 & 1.04 + 3 & + 0.0 & + 0.6 & @xmath132 & 0.56 + 4 & + 0.3 & + 0.0 & @xmath133 & 1.46 + 5 & + 0.3 & + 0.3 & @xmath134 & 1.14 + 6 & + 0.3 & + 0.6 & @xmath135 & 0.64 +
|
we discuss interstellar temperature determinations using the excitation equilibrium of the @xmath0 levels of si@xmath1ii and c@xmath1ii .
we show how observations of the @xmath2 fine structure levels of si@xmath1ii and c@xmath1ii ( which have significantly different excitation energies , corresponding to @xmath3 and 92 k , respectively ) can be used to limit gas kinetic temperatures .
we apply this method to the @xmath4 damped lyman-@xmath5 system toward the quasar pss 1443 + 27 . the lack of significant absorption out of the si@xmath1ii @xmath2 level and the presence of very strong c@xmath1ii @xmath2 provides an upper limit to the temperature of the c@xmath1ii@xmath6-bearing gas in this system . assuming a solar si / c ratio , the observations imply a @xmath7 limit @xmath8 k for this absorber ; a super - solar si / c ratio gives stricter limits , @xmath9 k. the observations suggest the presence of a cold neutral medium ; such cold gas may serve as the fuel for star formation in this young galaxy .
|
the discovery of the higgs boson ( @xmath0 ) , whose mass is 125 gev , at the large hadron collider ( lhc ) run - i @xcite and subsequent measurements of properties of the higgs boson @xcite have established the standard model ( sm ) of particle physics as a low energy effective theory at the electroweak scale .
nevertheless , the details of the higgs sector including the shape of the higgs potential are still unknown .
the higgs sector of the sm with one isospin doublet scalar field is constructed based merely on minimality , and not guided by any fundamental principle .
many models with extended higgs sectors can also satisfy current experimental data , and need to be distinguished in the future .
in addition to the ambiguity of the higgs potential , several phenomena that require physics beyond the sm ( bsm ) have been reported .
these include baryon asymmetry of the universe ( bau ) @xcite , the existence of dark matter , cosmic inflation and neutrino oscillations .
therefore , we are obliged to construct models of bsm to tackle such problems .
it should be noticed that many bsm models predict different classes of the higgs sector from the viewpoint of the number of higgs multiplets and imposed symmetries .
moreover , the essence of the higgs boson is directly connected to a bsm paradigm such as supersymmetry ( susy ) , dynamical symmetry breaking , classical scale invariance ( csi ) .
therefore , hints of new physics can be obtained by exploring the properties of the higgs sector .
one of the prominent examples of new physics accessible through probing the higgs potential is electroweak baryogenesis ( ewbg ) @xcite .
successful scenarios of ewbg require strongly first order phase transition ( 1stopt ) for the electroweak symmetry breaking ( ewsb ) . with the mass of the discovered higgs boson being @xmath1 ,
strongly 1stopt can not occur in the framework of the sm higgs sector .
however , by extending the higgs sector with additional scalar fields , strongly 1stopt can be easily realized @xcite . in order to unveil the mechanism of ewbg
, one needs to identify the shape of the higgs potential and the dynamics of the ewsb .
it is known that the condition for the 1stopt for ewsb leads to the deviation in the triple higgs boson coupling ( the @xmath2 coupling ) , which is typically larger than @xmath3 in renormalizable extended higgs sectors @xcite .
in addition , the deviation in the @xmath2 coupling and the 1stopt due to dimension - six operator in the higgs sector have been discussed in refs .
one method to investigate such strong 1stopt is to measure the @xmath2 coupling at future high - energy electron - positron colliders .
it seems to be challenging to measure the @xmath2 coupling with such an accuracy even at the high luminosity lhc .
once the international linear collider ( ilc ) @xcite is realized , the @xmath2 coupling can be measured with @xmath3 accuracy at the upgraded version with @xmath4 @xcite . the compact linear collider ( clic )
@xcite is also expected to be capable of testing the @xmath2 coupling at the similar accuracy , while the measurement of the @xmath2 coupling is beyond the scope of the future circular collider of electrons and positrons ( fcc - ee ) @xcite due to the lack of its center - of - mass energy . at future hadron colliders with the collision energy to be 100 tev , the similar accuracy to the ilc
may be expected for the measurement of the @xmath2 coupling as discussed in ref .
@xcite .
an alternative method to approach ewbg is observation of the stochastic gravitational waves ( gws ) produced by strongly first order electroweak phase transition ( ewpt ) @xcite . on february 11th , 2016 ,
the first direct detection of gws emitted by the merger of black holes at advanced ligo was reported @xcite .
it is expected that the ground - based experiments , kagra @xcite , advanced ligo @xcite and advanced virgo @xcite will soon observe many gw signals from binary systems of neutron stars or black holes .
thereby we have entered the new era of gw astronomy : physical phenomena can be probed through the observations of gws . although the main target of the above ground - based experiments is gws from astronomical phenomena , future space - based experiments such as elisa @xcite , decigo @xcite and bbo @xcite have the sensitivity to investigate some cosmological phenomena including strongly 1stopts and cosmic inflation at the early universe .
the spectrum of stochastic gws produced by strongly 1stopt has been investigated in the literature .
the gw amplitude and spectrum arising from dynamics of bubbles produced during the phase transition have been investigated in refs .
detectability of gws from strongly 1stopts at gw experiments has been discussed in refs .
ewpt and resulting gws in models where the higgs potential possesses higher dimensional operators have been studied in refs .
gw spectra from the strongly first order ewpt triggered by non - decoupling effects of new particles have been investigated in models with extended scalar sectors @xcite , and supersymmetric models @xcite . in the case of the minimal susy sm ( mssm )
, the required strongly 1stopt can not be realized according to negative results of stop searches at the lhc .
meanwhile , 1stopt and gws triggered by non - thermal effect at the tree level have been analyzed in the model with a singlet scalar extension @xcite , the left - right symmetric model @xcite , and the next - to - mssm ( nmssm ) @xcite .
gws in the scenario where ewbg is viable have been investigated in ref .
@xcite . in this paper
, we discuss complementarity between measurements of the @xmath2 coupling and the spectrum of gws in testing the strongly 1stopt of the electroweak symmetry in models with additional isospin singlet scalar fields with and without csi .
we demonstrate that the combination of the measurements of the @xmath2 coupling at future electron - positron colliders and gws at the space - based interferometers can probe physics behind the ewsb . as an example
, we consider a set of viable csi models with additional scalar fields where the ewsb is directly caused by thermal loop effects .
the idea of csi has been originally introduced by bardeen as a paradigm to avoid the hierarchy problem @xcite . in the csi models , utilizing the mechanism by coleman and weinberg @xcite , a mass scale is generated through the dimensional transmutation although any parameter with mass dimension is not introduced in the original lagrangian .
based on csi , many phenomenological models for ewsb have been proposed and investigated @xcite .
such a class of csi models can survive current experimental tests , and have the following unique features : a general upper bound on the lightest scalar boson is obtained ; the deviation in the higgs - photon - photon coupling is almost fixed by the number of charged scalar bosons ; and the deviation in the @xmath2 coupling is universally as large as @xmath5 @xcite .
these csi models also predict strongly 1stopt for ewpt . in ref .
@xcite , it is shown that the spectrum of gws from the ewpt strongly depends on the number of extra fields and their masses , and is useful in revealing the higgs potential and underlying physics .
we emphasize that the combination of future measurements of the @xmath2 coupling and stochastic gw signals would be a powerful tool in distinguishing the extended higgs sectors .
for concreteness , we focus on the csi models where @xmath6 extra isospin singlet scalars obey a global @xmath7 symmetry , and discuss how these models can be differentiated from similar extended models by means of the measurements of the @xmath2 coupling and the gw spectrum .
we find that they are synergetic in identifying specific models of ewsb in more details .
this paper is organized as follows . in sec .
[ subsec : massless ] , we briefly review the csi @xmath7 model .
the dynamics of 1stopt is discussed in sec .
[ sec:1stopt ] . in sec .
[ sec : gw ] , we discuss detectability of gw signals in the csi models based on new fitting functions provided in ref . @xcite .
we also compare the results in the csi models to those in the models equipped with mass parameters .
[ sec : conclusion ] is devoted to conclusion .
the landau pole in the csi @xmath7 model is discussed in appendix [ sec : landaupole ] , and gw signals evaluated by using the conventional fitting functions given in ref .
@xcite are summarized in appendix [ sec : comparison ] .
in this paper , we consider the following two types of @xmath7 scalar models as viable examples of strongly 1stopt , and compare both of them .
one is the models where additional isospin singlet scalar fields are added to the sm higgs sector with the mass square term which is supposed to be negative for ewsb . the @xmath2 coupling and the spectrum of gws have been discussed in these models in ref .
another type is the similar models but with csi . in this section
, we briefly review important phenomenological aspects of the @xmath7 scalar model with csi according to the work of ref .
@xcite .
the scalar sector consists of the higgs doublet @xmath8 and @xmath6 additional isospin singlet scalars @xmath9 @xcite . since the dynamics of the ewpt is basically controlled by the mass scale of additional particles and the number of these extra degrees of freedom , our results obtained are applicable to a wide class of csi models for ewsb .
imposing the csi , the scalar potential is given by symmetry of the csi models is exact , the extra scalars are stabilized and contribute to the relic abundance of the dark matter . applying the standard thermal dark matter production scenario to these csi models , the cases with @xmath10 are not favored in the light of dark matter direct detection experiments @xcite .
investigating phenomenology of dark matter is beyond the scope of this paper . ] @xmath11 as far as the csi is maintained , the vacuum expectation value ( vev ) of the higgs doublet as well as the mass of the higgs boson vanishes .
the csi is violated at the quantum level , and a mass scale is introduced due to the mechanism by coleman and weinberg @xcite .
the vacuum structure is investigated along the flat direction of the tree - level potential using the method developed by gildener and weinberg @xcite . introducing the order parameter @xmath12 along the flat direction as @xmath13 the one - loop effective potential at zero temperature is @xmath14 where @xmath15 is the renormalization scale . here ,
@xmath16 and @xmath17 are the field - dependent masses and the degrees of freedom of particles @xmath18 , respectively .
the degrees of freedom are given by @xmath19 we take the @xmath20 scheme @xcite , where @xmath21 irrespective of particle spins .
the one - loop contribution from the higgs boson is absent as this effect arises from higher - order corrections . in the csi models ,
the field - dependent mass squared of the field @xmath18 is proportional to @xmath22 as @xmath23 .
the effective potential is rewritten as @xmath24 where @xmath25 using the stationary condition , we have @xmath26 where the renormalization scale @xmath15 is fixed .
the mass squared of the discovered higgs boson is obtained as @xmath27 then , the one - loop effective potential is given in terms of the renormalized mass of the higgs boson as @xmath28 the renormalized @xmath2 coupling is obtained as @xmath29 where @xmath30 is the tree - level @xmath2 coupling in the sm .
then , the deviation in the @xmath2 coupling is defined by coupling , there is a case where the deficit caused by the top loop contribution obtained through the effective potential method is compensated @xcite .
accordingly , we take the tree level value for the sm prediction of the @xmath2 coupling . ]
@xmath31 independently of the details of the models , the csi models always predict @xmath32 . such a characteristic feature of the csi models for the deviation in the @xmath2 coupling is expected to be tested at future collider experiments by measuring the double higgs boson production processes . at the high luminosity lhc with the integrated luminosity of @xmath33 ,
the production cross section can be measured with @xmath34 @xcite , but a measurement of the @xmath2 coupling with the desired accuracy would be challenging . at the ilc with the center - of - mass energy of @xmath35 and the luminosity of @xmath36 , the @xmath2 coupling is expected to be determined at a precision of @xmath37 @xcite .
the expected accuracy on the @xmath2 coupling at the ilc stage with @xmath4 and @xmath38 ( @xmath39 ) is @xmath40 ( @xmath3 ) @xcite .
therefore , this class of models can be tested at the ilc ( and also the clic ) .
coupling with the similar accuracy @xcite . ]
in the csi @xmath7 models , the effective potential at finite temperatures is given at the one - loop level by @xcite @xmath41 the finite - temperature contribution to the effective potential is written as @xmath42 where @xmath43 , \quad a_i = m_i(\varphi , t)/t .
\end{aligned}\ ] ] here , we employ a ring - improved effective potential that is obtained by replacing the field - dependent masses in eq .
( [ eq : v1 ] ) by @xmath44 where @xmath45 is the finite temperature contribution to the self - energies @xcite .
the thermally corrected field - dependent masses of the electroweak gauge bosons are @xmath46 in the @xmath47 basis with @xmath48 , @xmath49 .
notice that only the self - energy for the longitudinal modes of the gauge bosons receive thermal corrections .
the field - dependent masses of fermions do not receive thermal corrections , @xmath50 the thermally corrected field - dependent mass of the singlet scalars are explained by @xmath51 where @xmath52.\end{aligned}\ ] ] in order to generate bau , sakharov specified three necessary conditions @xcite .
one of them is to require the departure from thermal equilibrium , which is accomplished if the baryon number changing sphaleron interaction quickly decouples after the ewsb .
this criterion for the sphaleron decoupling is described by @xmath53 where @xmath54 is the sphaleron interaction rate and @xmath55 is the hubble parameter at @xmath56 .
this condition requires that the ewsb must be of strongly first order @xcite , @xmath57 where @xmath58 is the vev for the broken phase minimum at the critical temperature @xmath59 . in this paper
, we numerically compute @xmath58 and @xmath59 instead of using high temperature expansion . in fig .
[ fig : pt_masslesso(n ) ] , we show contours of the mass of the higgs boson @xmath60 and @xmath61 in the @xmath62 plane in the csi @xmath7 model . with the mass of the higgs boson being 125 gev ,
strongly 1stopts of @xmath63 are realized for @xmath64 .
it is characteristic of the csi models to satisfy the condition for strongly 1stopt for the ewsb .
the right - top shaded region is excluded by the perturbative unitarity bound @xcite . in order to keep perturbative unitarity ,
we impose @xmath65 for the eigenvalues of @xmath66-wave amplitudes of two body elastic scatterings among longitudinally polarized weak bosons and scalar bosons .
we find that it is sufficient to impose the following inequality : @xmath67<\frac{1}{2}. \label{pu}\end{aligned}\ ] ] in fig .
[ fig : pt_masslesso(n ) ] , vanishingly small @xmath68 and @xmath69 are assumed as these parameters do not affect the computation of the gw spectrum discussed in the next section . and @xmath61 in the @xmath62 plane in the csi @xmath7 model.,scaledwidth=48.0% ]
we here introduce two important quantities @xmath70 and @xmath71 which parametrize the spectrum of gws following ref . @xcite .
one of the key quantities that describe dynamics of vacuum bubble is the bubble nucleation rate per unit volume per unit time , @xmath72 , \end{aligned}\ ] ] where @xmath73 is the euclidean action of a critical bubble and @xmath74 . to parametrize the rate of variation of the bubble nucleation rate at the time of the phase transition @xmath75
, we introduce the quantity @xmath71 defined as @xmath76 the inverse of the parameter @xmath71 describes approximately the time duration of the phase transition . at finite temperatures
, the @xmath77-symmetric field configuration @xmath78 contributes to the action as @xmath79 , where @xmath80 is the three dimensional euclidean action , @xmath81.\end{aligned}\ ] ] the bounce solution @xmath82 is obtained by solving the differential equation , @xmath83 with the boundary conditions @xmath84 it is convenient to introduce the dimensionless parameter @xmath85 in order for the nucleated vacuum bubbles to percolate through the universe , the nucleation rate per hubble volume per hubble time should reach the unity @xmath86 this condition is converted to the relation @xmath87 another important quantity is the false - vacuum energy density available for the evolution of bubbles @xmath88 where @xmath89 is the vev for the broken phase minimum at @xmath56 , and we reset the symmetric phase minimum to zero .
we introduce the parameter @xmath70 parametrizing the ratio of the false - vacuum energy density @xmath90 to the thermal energy density @xmath91 in the symmetric phase at the transition temperature @xmath92 , @xmath93 where the radiation energy density is given by @xmath94 , with @xmath95 being the relativistic degrees of freedom in the thermal plasma .
the predictions for @xmath70 , @xmath96 , @xmath97 , @xmath61 , @xmath59 and @xmath92 in the csi @xmath7 models are shown in the top frame of table [ tab : on ] . in the middle and bottom frames of table [
tab : on ] , for comparison , we also listed the predictions of the @xmath7 models with mass terms that produce @xmath98 for @xmath99 and @xmath100 , respectively .
( for the details of the @xmath7 models with mass terms ; i.e. the @xmath7 models without csi , see ref .
@xcite ) . in the @xmath7 models without csi , @xmath61 and @xmath70 ( @xmath96 ) decrease ( increases ) with increasing @xmath101 for a fixed value of @xmath102 . in fig .
[ fig : alphabeta ] , the predicted values of @xmath70 and @xmath71 in the csi @xmath7 models ( red ) and @xmath7 models without csi for @xmath99 ( gray ) are plotted . due to the finite - temperature effect ,
the csi is violated . as a result ,
the predicted values of @xmath70 and @xmath71 as well as @xmath61 significantly change depending on the number of the extra scalar bosons although the @xmath2 coupling is common in the csi models .
the upper ( lower ) bound on @xmath71 for each @xmath7 model without csi is derived by the condition of @xmath103 ( @xmath104 ) .
the predicted values of @xmath70 and @xmath71 strongly depend on the number of extra scalars @xmath6 and their mass @xmath105 as pointed out in ref .
this fact opens new possibilities for resolving these models through observation of gw signals . and @xmath96 in the csi @xmath7 models ( red ) and @xmath7 models without csi for @xmath99 ( gray).,scaledwidth=48.0% ] during the 1stopt in the early universe , collisions of vacuum bubbles and bulk motion of the plasma produce gws .
there are three contributions to the stochastic gw background from the 1stopt ; collisions of bubble walls , the compression waves in the plasma ( sound waves ) and magnetohydrodynamic ( mhd ) turbulence in the plasma .
computation of the produced spectrum of the gws from the 1stopt depends on the details of dynamics of colliding bubbles , and complicated numerical simulations are required . here
, we employ the fitting functions recently provided in ref .
@xcite for estimating the gw spectrum from the parameters @xmath70 and @xmath71 . in our previous work in the @xmath7 models without csi
@xcite , we employed the fitting function for gws from bubble collisions in the envelope approximation in ref .
@xcite and that for gws from plasma turbulence in ref .
@xcite . in appendix
[ sec : comparison ] , we discuss differences between these fitting functions and those in ref .
@xcite .
the gw spectrum from the collision of bubble walls can be computed using the method referred to as the envelope approximation @xcite .
a fit to the spectrum obtained by numerical simulations is given by @xmath106 where the peak of the energy density is given by @xmath107 at the peak frequency @xmath108 here , @xmath109 is the wall velocity of the bubble , and the parameter @xmath110 is the ratio of the vacuum energy transferred to the bubble motion .
the relativistic degrees of freedom at the transition temperature is denoted by @xmath111 . as for the contribution to gws from sound waves , a fitting function to the numerical simulations is obtained as @xcite @xmath112 where the peak energy density is given by @xmath113 at the peak frequency @xmath114 in addition to the compression waves , turbulent motion in the plasma also contributes to the gw spectrum .
taking mhd effects on the ionized plasma into account , the gw spectrum from the mhd turbulence is @xcite @xmath115 where @xmath116 and @xmath117 the mhd turbulent contribution depends on the hubble parameter at the time of gw production , whose redshifted value is @xmath118 according to the recent simulations @xcite , the fraction of the turbulent bulk motion is at most @xmath119 . in our later numerical analysis ,
we set @xmath120 .
the efficiency factor @xmath121 denotes the fraction of the vacuum energy that is transformed into the bulk motion of the plasma fluid , and given as a function of @xmath70 and @xmath109 .
we refer to results obtained in ref .
@xcite , @xmath122 \quad ( \text{for}~ c_s < v_b < v_j ) \\ & \frac { ( v_j - 1)^3 v_j^{5/2 } v_b^{-5/2 } \kappa_c \kappa_d } { [ ( v_j -1)^3 - ( v_b-1)^3 ] v_j^{5/2 } \kappa_c + ( v_b - 1)^3 \kappa_d } \quad ( \text{for}~ v_j \lesssim v_b ) \end{cases } , \end{aligned}\ ] ] where @xmath123 the efficiency factor is treated separately depending on whether the wall velocity @xmath124 exceeds the velocity of sound ( @xmath125 ) or @xmath126 the derivative of @xmath121 with respect to @xmath109 at @xmath127 is approximately given by @xmath128 the dependence of the efficiency factor @xmath121 on @xmath109 and @xmath70 is shown in fig .
[ fig : kappa ] . for large @xmath70 and small @xmath109 ,
no consistent solution to hydrodynamic equations exists @xcite . as a function of @xmath124 for @xmath129 and @xmath130 .
the cases for @xmath131 and @xmath132 are plotted with dotted lines . , scaledwidth=48.0% ] if the vacuum energy released at the phase transition is adequately large , it is possible for the bubble wall velocity to approach the speed of light @xcite .
the minimum value of @xmath70 for realizing this runaway bubble wall ( @xmath133 ) is approximately @xcite @xmath134,\end{aligned}\ ] ] where @xmath135 ( @xmath136 ) for bosons ( fermions ) .
particles that are light in the symmetric phase and become non - relativistic after the phase transition contribute to the sum on the right - hand side of the above equation . in the runaway case , abundant energy can be used for accelerating the bubble wall , enhancing the contribution from the wall collision to gws .
the efficiency factors are given by @xmath137 where @xmath138 .
in the non - runaway case ( @xmath139 ) , on the other hand , the wall collision contribution is small @xcite .
we here show our numerical results of the spectrum of gws from strongly 1stopt in the @xmath7 model with and without csi .
we then discuss the detectability of the gws at the future space - based interferometers , elisa and decigo . in fig .
[ fig : gwspectrum ] ( top frames ) , we show the predicted gw spectra produced from the ewpt in the csi @xmath7 models using the formulae provided above .
the black curves correspond to the contributions from the sound waves ( solid ) and turbulence ( dashed ) for @xmath140 and @xmath141 from the bottom .
see table [ tab : on ] for the predicted values .
the csi @xmath7 model with @xmath142 requires @xmath143 to obtain a consistent solution to hydrodynamics equations , and is not displayed in the right frame .
it turns out that the contribution from sound waves tends to dominate .
the colored regions show experimental sensitivities expected at future space - based interferometers , elisa @xcite and decigo @xcite . the labels `` c1 '' , `` c2 '' , `` c3 '' and `` c4 '' correspond to four different configurations of elisa provided in table 1 in ref .
@xcite whereas the labels `` pre - decigo '' , `` fp - decigo '' and `` correlation '' are decigo designs @xcite . as for the wall velocity ,
we take two reference values @xmath144 ( left ) and @xmath145 ( right ) . for @xmath144 ,
even the elisa c4 design has sensitivity to observe the predicted gw spectra in the csi model with @xmath142 .
the elisa c1 design can probe the csi models with @xmath146 .
the energy density of gws is suppressed and its peak frequency is enhanced for @xmath145 . in this case , decigo will be sensitive even to the model with @xmath147 taking the correlation of nearby two clusters . for smaller values of @xmath124 ,
the generated gws are too small to detect at the elisa c1 design . for successful ewbg
, however , subsonic wall velocities as small as @xmath148 are generically required @xcite .
contrary to this argument , ref .
@xcite points out there are some cases where ewbg is possible even if wall velocities are supersonic and gw signals are large .
the upper bound on the gws is derived from non - observation of extra radiation @xcite , and indicated by @xmath149 .
we comment on foreground noise that may obscure the gw signals from the ewpt .
the maximum value of the contribution from extragalactic white dwarf binaries is estimated to be @xmath150 at @xmath151 @xcite , while that from galactic white dwarf binaries is significant for @xmath152 hz @xcite .
therefore , the sound wave contribution to gws spectrum is not screened by these astrophysical sources for large @xmath6 and large wall velocities . for comparison ,
the gw spectra predicted in the @xmath7 models without csi that coincidentally leads to @xmath98 are shown in fig .
[ fig : gwspectrum ] ( bottom frames ) . the @xmath7 model without csi with @xmath142
requires @xmath153 , and is not displayed in the right frame .
models ( top frames ) and @xmath7 models without csi with @xmath98 ( bottom frames ) for @xmath144 ( left ) and @xmath154 ( right ) .
the black curves correspond to the contributions from the sound waves ( solid ) and turbulence ( dashed ) for @xmath140 and @xmath141 ( left ) for @xmath155 and @xmath156 ( right ) from the bottom .
, scaledwidth=100.0% ] models ( top frames ) and @xmath7 models without csi with @xmath98 ( bottom frames ) for @xmath144 ( left ) and @xmath154 ( right ) .
the black curves correspond to the contributions from the sound waves ( solid ) and turbulence ( dashed ) for @xmath140 and @xmath141 ( left ) for @xmath155 and @xmath156 ( right ) from the bottom .
, scaledwidth=100.0% ] models ( top frames ) and @xmath7 models without csi with @xmath98 ( bottom frames ) for @xmath144 ( left ) and @xmath154 ( right ) .
the black curves correspond to the contributions from the sound waves ( solid ) and turbulence ( dashed ) for @xmath140 and @xmath141 ( left ) for @xmath155 and @xmath156 ( right ) from the bottom .
, scaledwidth=100.0% ] models ( top frames ) and @xmath7 models without csi with @xmath98 ( bottom frames ) for @xmath144 ( left ) and @xmath154 ( right ) .
the black curves correspond to the contributions from the sound waves ( solid ) and turbulence ( dashed ) for @xmath140 and @xmath141 ( left ) for @xmath155 and @xmath156 ( right ) from the bottom .
, scaledwidth=100.0% ] the gw spectra for the runaway case ( @xmath157 ) are shown in fig .
[ fig : gwspectrum2 ] .
the runaway bubble wall can be realized for @xmath158 and @xmath141 in the @xmath7 models without csi with @xmath100 among our benchmark points . in the csi @xmath7 models and other @xmath7 models without csi ,
a large number of singlet scalars become heavy during the phase transition and contribute to the friction on the bubble wall , preventing the wall from the runaway behavior . models without csi with @xmath100 and @xmath98 for the runaway case ( @xmath157 ) .
the black curves correspond to the contributions from the sound waves ( solid ) , bubble collision ( dotted ) and turbulence ( dashed ) for @xmath159 and @xmath141 from the bottom . , scaledwidth=48.0% ] in the case where the deviation in the @xmath2 coupling is found to be around @xmath5 at electron - positron collider , it is challenging for collider experiments to distinguish the csi models from other extended models . in fig .
[ fig : alphabeta_massless ] , the detectability of gws in the csi @xmath7 models ( red ) and @xmath7 model without csi with @xmath98 ( gray ) are displayed .
the experimental sensitivities expected at the several designs of elisa and decigo are set by using the sound wave contribution .
actually , the transition temperature @xmath160 differs according to models as presented in table [ tab : on ] . for the purpose of illustration ,
the transition temperature is fixed at @xmath161 ( top ) and @xmath162 ( bottom ) . as in fig .
[ fig : gwspectrum ] , we take @xmath144 ( left ) and @xmath145 ( right ) . since the shape of the effective potentials of the csi models is different from those of the extended models without csi , their predicted regions in the ( @xmath70 , @xmath71 ) plane do not overlap .
if the wall velocity is as large as @xmath144 , elisa c1 configuration can detect the produced gws in the csi models with @xmath146 .
the sensitivity reach of the correlation program of decigo can cover all the csi models even when the wall velocity is as small as @xmath145 .
therefore , the two classes of model can be differentiated at future gw detectors even if they share the common value for the @xmath2 coupling .
models ( red ) and @xmath7 models without csi with @xmath98 ( gray ) .
the experimental sensitivities expected at the several designs of elisa and decigo are set by using the sound wave contribution for @xmath161 ( top frames ) and @xmath162 ( bottom frames ) .
we take @xmath144 ( left ) and @xmath145 ( right ) .
the upper bound on @xmath70 ( @xmath163 ) is delineated for @xmath164 ( right ) .
, scaledwidth=100.0% ] models ( red ) and @xmath7 models without csi with @xmath98 ( gray ) .
the experimental sensitivities expected at the several designs of elisa and decigo are set by using the sound wave contribution for @xmath161 ( top frames ) and @xmath162 ( bottom frames ) .
we take @xmath144 ( left ) and @xmath145 ( right ) .
the upper bound on @xmath70 ( @xmath163 ) is delineated for @xmath164 ( right ) .
, scaledwidth=100.0% ] models ( red ) and @xmath7 models without csi with @xmath98 ( gray ) .
the experimental sensitivities expected at the several designs of elisa and decigo are set by using the sound wave contribution for @xmath161 ( top frames ) and @xmath162 ( bottom frames ) .
we take @xmath144 ( left ) and @xmath145 ( right ) .
the upper bound on @xmath70 ( @xmath163 ) is delineated for @xmath164 ( right ) .
, scaledwidth=100.0% ] models ( red ) and @xmath7 models without csi with @xmath98 ( gray ) .
the experimental sensitivities expected at the several designs of elisa and decigo are set by using the sound wave contribution for @xmath161 ( top frames ) and @xmath162 ( bottom frames ) .
we take @xmath144 ( left ) and @xmath145 ( right ) .
the upper bound on @xmath70 ( @xmath163 ) is delineated for @xmath164 ( right ) .
, scaledwidth=100.0% ] before closing this section , we mention the predictions in the csi two higgs doublet model ( 2hdm ) @xcite . the higgs sector of the 2hdm consists of two higgs doublets @xmath165 and @xmath166
. therefore , the csi 2hdm is an explicit example corresponding to the csi @xmath167 model as the number of the extra scalars is common .
we have numerically checked that the predicted values of @xmath70 and @xmath71 are approximately equal to those in the csi @xmath167 model .
nature of ewpt is important for understanding physics at the early universe , such that the strongly 1stopt is required for a successful scenario of ewbg .
we have discussed complementarity of the methods of measurements of the @xmath2 coupling at the ilc and the spectrum of stochastic gws at elisa and decigo in testing the strongly 1stopt of the ewsb in the @xmath7 models with and without csi .
the models with csi universally predict a large deviation in the @xmath2 coupling as large as about @xmath5 , and leads to the strongly first order ewpt .
there is a possibility that the deviation in the @xmath2 coupling in the @xmath7 models without csi accidentally matches @xmath5 .
even in such a case , the predicted gw spectra are different with each other , and we can separate both the models by measuring gws at future space - based interferometers . in conclusion ,
synergy between the future measurements of the @xmath2 coupling and the gw signals provides us important hints regarding narrowing down the dynamics behind the ewsb .
this work was supported , in part , by grant - in - aid for scientific research , no .
26104702 ( mk ) and no . 23104006 ( sk ) , grant h2020-msca - rise-2014
645722 ( non minimal higgs ) ( sk ) , and the sasakawa scientific research grant from the japan science society ( tm ) .
we comment on the occurrence of the landau pole in the csi models .
although values of the couplings in the scalar sector are in the perturbative range at around the electroweak scale , large scalar couplings lead to the landau pole at energy scales near the electroweak scale .
therefore , some theory of uv completion should replace the csi models above this cutoff scale @xmath168 .
let us estimate the energy scale @xmath168 of the landau pole in the csi @xmath7 models by solving the renormalization group equations .
we derived the beta functions at one - loop level which are consistent with the ones in ref .
@xcite . as for the input values of the couplings in the scalar sector , we take @xmath169 and @xmath170 at the boundary scale @xmath15 fixed by eq .
( [ eq : q ] ) .
the value of the singlet self - coupling @xmath69 is undetermined and constrained only by the unitarity bound . in table
[ tab : landau ] , the energy scale of the the landau pole @xmath168 is shown for @xmath171 and @xmath141 . following the convention adopted in ref .
@xcite , the cutoff scale @xmath168 in table [ tab : landau ] is defined as the scale where any of the scalar couplings diverges . to keep perturbative expansion . in view of uncertainties involved in the analysis of the renormalization group flow
, we disregard this discrepancy . ] for @xmath172 , the csi model with @xmath142 satisfying the observed higgs boson mass @xmath173 is excluded by the unitarity bound . as shown in the table , typically , the landau pole appears at around @xmath174 tev .
as @xmath69 is increased , the location of the landau pole is shifted to lower energy scales . for moderate values of @xmath175 , destructive interference between
the top yukawa and @xmath176 contributions is large , leading to higher cutoff scales .
.the energy scale of the landau pole @xmath168 in the csi @xmath7 models for @xmath171 and @xmath141 . [ cols="<,^,^,^,^",options="header " , ]
in fig . [ fig : alphabeta_new ] , we present the experimental sensitivity to the contributions from mhd turbulence in addition to the sound waves at @xmath161 . as illustrated , we see that the sound waves are the dominant contribution . next , we compare these results based on the recent work given in ref .
@xcite with the estimate of the gw spectrum based on ref .
@xcite which we employed in our previous work @xcite .
results based on the conventional analysis given in ref .
@xcite is shown in fig .
[ fig : alphabeta_old ] . as for the contribution from the turbulence in the plasma , the difference between fig .
[ fig : alphabeta_new ] ( right ) and fig .
[ fig : alphabeta_old ] ( right ) is attributed to mhd effects @xcite . plane based on the conventional analysis given in ref .
we show the contributions from the collision of bubble walls using the envelope approximation ( left ) and turbulence ( right ) assuming @xmath161 .
, scaledwidth=100.0% ] plane based on the conventional analysis given in ref . @xcite .
we show the contributions from the collision of bubble walls using the envelope approximation ( left ) and turbulence ( right ) assuming @xmath161 .
, scaledwidth=100.0% ] in fig .
[ fig : gwspectrum_old ] , we show gw spectra based on the conventional analysis given in ref .
@xcite , since the sound wave contribution is absent , the predicted gws are relatively weaker compared to those shown in fig .
[ fig : gwspectrum ] . models ( left ) and the @xmath7 models without csi with @xmath98 ( right ) based on the conventional analysis given in ref .
the black curves correspond to the contributions from the bubble collision ( dotted ) and turbulence ( dashed ) for @xmath140 and @xmath141 from the bottom .
, scaledwidth=100.0% ] models ( left ) and the @xmath7 models without csi with @xmath98 ( right ) based on the conventional analysis given in ref .
the black curves correspond to the contributions from the bubble collision ( dotted ) and turbulence ( dashed ) for @xmath140 and @xmath141 from the bottom .
, scaledwidth=100.0% ] 99 g. aad _ et al . _ [ atlas collaboration ] , phys .
b * 716 * , 1 ( 2012 ) .
s. chatrchyan _ et al . _
[ cms collaboration ] , phys .
b * 716 * , 30 ( 2012 ) .
v. khachatryan
_ et al . _
[ cms collaboration ] , eur .
j. c * 75 * , no . 5 , 212 ( 2015 ) . g. aad _ et al .
_ [ atlas collaboration ] , eur .
j. c * 76 * , no . 1 , 6 ( 2016 ) . k. a. olive _ et al .
_ [ particle data group collaboration ] , chin .
c * 38 * , 090001 ( 2014 ) .
v. a. kuzmin , v. a. rubakov and m. e. shaposhnikov , phys .
b * 155 * , 36 ( 1985 ) .
k. funakubo , a. kakuto and k. takenaga , prog .
phys . * 91 * , 341 ( 1994 ) . j. m. cline and p. a. lemieux , phys .
d * 55 * , 3873 ( 1997 ) .
s. kanemura , y. okada and e. senaha , phys .
b * 606 * , 361 ( 2005 ) .
k. funakubo , s. tao and f. toyoda , prog .
theor . phys .
* 114 * , 369 ( 2005 ) .
s. profumo , m. j. ramsey - musolf and g. shaughnessy , jhep * 0708 * , 010 ( 2007 ) .
a. noble and m. perelstein , phys .
d * 78 * , 063518 ( 2008 ) .
m. aoki , s. kanemura and o. seto , phys .
lett . * 102 * , 051805 ( 2009 ) .
s. kanemura , e. senaha and t. shindou , phys .
b * 706 * , 40 ( 2011 ) .
g. gil , p. chankowski and m. krawczyk , phys .
b * 717 * , 396 ( 2012 ) . c. tamarit , phys
. rev .
d * 90 * , no . 5 , 055024 ( 2014 ) . s. kanemura , n. machida and t. shindou , phys .
b * 738 * , 178 ( 2014 ) .
k. fuyuto and e. senaha , phys . rev .
d * 90 * , no .
1 , 015015 ( 2014 ) . s. profumo , m. j. ramsey - musolf , c. l. wainwright and p. winslow , phys .
d * 91 * , no .
3 , 035018 ( 2015 ) .
n. blinov , s. profumo and t. stefaniak , jcap * 1507 * , no .
07 , 028 ( 2015 ) .
k. fuyuto and e. senaha , phys .
b * 747 * , 152 ( 2015 ) .
a. karam and k. tamvakis , phys .
d * 92 * , no . 7 , 075010 ( 2015 ) .
m. kakizaki , s. kanemura and t. matsui , phys .
d * 92 * , no .
11 , 115007 ( 2015 ) . c. grojean , g. servant and j. d. wells , phys .
d * 71 * , 036001 ( 2005 ) .
b. grinstein and m. trott , phys .
d * 78 * , 075022 ( 2008 ) .
j. brau , ( ed . )
_ et al . _
[ ilc collaboration ] , arxiv:0712.1950 [ physics.acc-ph ] ; g. aarons _ et al . _
[ ilc collaboration ] , arxiv:0709.1893 [ hep - ph ] ; n. phinney , n. toge and n. walker , arxiv:0712.2361 [ physics.acc-ph ] ; t. behnke , ( ed . ) _ et al . _ [ ilc collaboration ] , arxiv:0712.2356 [ physics.ins-det ] ; t. behnke _ et al . _ , arxiv:1306.6329 [ physics.ins-det ] ; h. baer , _ et al .
_ `` physics at the international linear collider '' , _ physics chapter of the ilc detailed baseline design report _ : http://lcsim.org/papers/dbdphysics.pdf .
d. m. asner , t. barklow , c. calancha , k. fujii , n. graf , h. e. haber , a. ishikawa and s. kanemura _ et al .
_ , arxiv:1310.0763 [ hep - ph ] .
k. fujii _ et al .
_ , arxiv:1506.05992 [ hep - ex ] .
e. accomando _ et al .
_ [ clic physics working group collaboration ] , hep - ph/0412251 ; l. linssen , a. miyamoto , m. stanitzki and h. weerts , arxiv:1202.5940 [ physics.ins-det ] .
m. bicer _ et al . _
[ tlep design study working group collaboration ] , jhep * 1401 * , 164 ( 2014 ) .
h. j. he , j. ren and w. yao , phys . rev .
d * 93 * , no . 1 , 015003 ( 2016 ) .
c. grojean and g. servant , phys .
d * 75 * , 043507 ( 2007 ) .
b. p. abbott _ et al . _ [ ligo scientific and virgo collaborations ] , phys .
lett . * 116 * , no . 6 , 061102 ( 2016 ) .
k. somiya [ kagra collaboration ] , class .
* 29 * , 124007 ( 2012 ) .
g. m. harry [ ligo scientific collaboration ] , class .
* 27 * , 084006 ( 2010 ) .
t. accadia _ et al .
_ , proceedings of _ 12th marcel grossmann meeting on general relativity _ , pp .
17381742 ( 2009 ) .
a. seoane _ et al .
_ [ elisa collaboration ] , arxiv:1305.5720 [ astro-ph.co ] .
s. kawamura _ et al .
_ , class .
* 28 * , 094011 ( 2011 ) . v. corbin and n. j. cornish , class .
* 23 * , 2435 ( 2006 ) .
m. kamionkowski , a. kosowsky and m. s. turner , phys .
d * 49 * , 2837 ( 1994 ) .
s. j. huber and t. konstandin , jcap * 0809 * , 022 ( 2008 ) .
j. r. espinosa , t. konstandin , j. m. no and g. servant , jcap * 1006 * , 028 ( 2010 ) .
j. m. no , phys .
d * 84 * , 124025 ( 2011 ) . m. hindmarsh , s. j. huber , k. rummukainen and d. j. weir , phys .
lett . * 112 * , 041301 ( 2014 ) .
m. hindmarsh , s. j. huber , k. rummukainen and d. j. weir , arxiv:1504.03291 [ astro-ph.co ] . c. caprini _
_ , arxiv:1512.06239 [ astro-ph.co ] .
l. leitao , a. megevand and a. d. sanchez , jcap * 1210 * , 024 ( 2012 ) .
y. kikuta , k. kohri and e.
so , arxiv:1405.4166 [ hep - ph ] .
r. jinno , k. nakayama and m. takimoto , phys .
d * 93 * , no . 4 , 045024 ( 2016 ) .
l. leitao and a. megevand , arxiv:1512.08962 [ astro-ph.co ] .
j. jaeckel , v. v. khoze and m. spannowsky , arxiv:1602.03901 [ hep - ph ] .
p. s. b. dev and a. mazumdar , arxiv:1602.04203 [ hep - ph ] . c. delaunay , c. grojean and j. d. wells , jhep * 0804 * , 029 ( 2008 ) .
huang , y. wan , d. g. wang , y. f. cai and x. zhang , arxiv:1601.01640 [ hep - ph ] .
j. r. espinosa and m. quiros , phys .
d * 76 * , 076004 ( 2007 ) .
j. r. espinosa , t. konstandin , j. m. no and m. quiros , phys .
d * 78 * , 123528 ( 2008 ) .
r. apreda , m. maggiore , a. nicolis and a. riotto , nucl .
b * 631 * , 342 ( 2002 ) .
a. ashoorioon and t. konstandin , jhep * 0907 * , 086 ( 2009 ) .
l. sagunski , desy - thesis-2013 - 011 .
s. j. huber , t. konstandin , g. nardini and i. rues , arxiv:1512.06357 [ hep - ph ] . w. a. bardeen , fermilab - conf-95 - 391-t .
s. r. coleman and e. j. weinberg , phys .
d * 7 * , 1888 ( 1973 ) .
e. gildener and s. weinberg , phys .
d * 13 * , 3333 ( 1976 ) .
k. takenaga , prog .
phys . * 92 * , 987 ( 1994 ) ; * 95 * , 609 ( 1996 ) .
j. s. lee and a. pilaftsis , phys .
d * 86 * , 035004 ( 2012 ) .
k. endo and y. sumino , jhep * 1505 * , 030 ( 2015 ) .
k. endo and k. ishiwata , phys .
b * 749 * , 583 ( 2015 ) .
a. d. plascencia , jhep * 1509 * , 026 ( 2015 ) .
k. hashino , s. kanemura and y. orikasa , phys .
b * 752 * , 217 ( 2016 ) .
k. endo , k. ishiwata and y. sumino , arxiv:1601.00696 [ hep - ph ] . w. siegel , phys .
b * 84 * , 193 ( 1979 ) ; * 94 * , 37 ( 1980 ) ; d. m. capper , d. r. t. jones and p. van nieuwenhuizen , nucl . phys .
b * 167 * , 479 ( 1980 ) .
s. kanemura , s. kiyoura , y. okada , e. senaha and c. p. yuan , phys .
b * 558 * , 157 ( 2003 ) .
s. kanemura , y. okada , e. senaha and c .- p .
yuan , phys .
d * 70 * , 115002 ( 2004 ) .
cms collaboration [ cms collaboration ] , cms - pas - ftr-15 - 002 .
l. dolan and r. jackiw , phys .
d * 9 * , 3320 ( 1974 ) .
m. e. carrington , phys .
d * 45 * , 2933 ( 1992 ) .
a. d. sakharov , pisma zh .
* 5 * , 32 ( 1967 ) . c. caprini , r. durrer and g. servant , jcap * 0912 * , 024 ( 2009 ) .
a. nicolis , class .
* 21 * , l27 ( 2004 ) .
p. binetruy , a. bohe , c. caprini and j. f. dufaux , jcap * 1206 * , 027 ( 2012 ) .
a. klein _ et al .
_ , phys .
d * 93 * , no . 2 , 024003 ( 2016 ) .
data sheet by a. petiteau , + http://www.apc.univ-paris7.fr/downloads/lisa/elisa/sensitivity/cfgv1/stochbkgd/ m. joyce , t. prokopec and n. turok , phys .
lett . * 75 * , 1695 ( 1995 ) ; phys .
d * 53 * , 2930 ( 1996 ) ; * 53 * , 2958 ( 1996 ) .
p. a. r. ade _ et al .
_ [ planck collaboration ] , arxiv:1502.01589 [ astro-ph.co ] .
r. schneider , s. marassi and v. ferrari , class .
* 27 * , 194007 ( 2010 ) .
g. nelemans , class .
* 26 * , 094030 ( 2009 ) .
s. j. huber and t. konstandin , jcap * 0805 * , 017 ( 2008 ) .
|
probing the higgs potential and new physics behind the electroweak symmetry breaking is one of the most important issues of particle physics .
in particular , nature of electroweak phase transition is essential for understanding physics at the early universe , such that the strongly first order phase transition is required for a successful scenario of electroweak baryogenesis . the strongly first order phase transition is expected to be tested by precisely measuring the triple higgs boson coupling at future colliders like the international linear collider .
it can also be explored via the spectrum of stochastic gravitational waves to be measured at future space - based interferometers such as elisa and decigo .
we discuss complementarity of both the methods in testing the strongly first order phase transition of the electroweak symmetry in models with additional isospin singlet scalar fields with and without classical scale invariance .
we find that they are synergetic in identifying specific models of electroweak symmetry breaking in more details .
|
the study of the evolution of small perturbations in black hole backgrounds is a very interesting subject . actually , we know that this evolution , at intermediate times , is dominated by damped single frequency oscillations .
these characteristic oscillations have been termed quasinormal modes and the associated frequencies quasinormal frequencies , and depend only on the parameters characterizing the black hole , as it mass , electric charge and angular momentum @xcite . in this sense
we can say that black holes have a characteristic sound , resembling for example the familiar sound produced bay the ringing of a bell or the strum of a guitar .
once one realizes the importance of black holes in fundamental physics , one can grasp the meaning and significance of their characteristic vibrations @xcite .
black holes have been called the hydrogen atom of general relativity , perfect comparison because like the hydrogen atom in quantum mechanics , a black hole , as a solution of the einstein s field equations has all the general relativistic properties embodied in it , but still is simple enough to be a model for starting a complete understanding of all the physics that go with einstein s gravitation theory .
there are different contexts in which the study of quasinormal modes of black holes appears to be motivated : the estimation of astrophysical black hole parameters @xcite , the study of the stability of these solutions under small perturbations , semiclassical ways to quantize the black hole area and the estimation of thermalization timescales in connection with the ads / cft correspondence , in which a large static black hole in asymptotically ads spacetime corresponds to a thermal state in a conformal field theory ( cft ) at its boundary , and the decay of the test field in the black hole spacetime corresponds to the decay of the perturbed state in the cft . in this sense
, we can determine the timescale for the return to thermal equilibrium studying its dynamics in ads spactime , and then translating it onto the cft , using the ads / cft conjecture @xcite @xcite .
since the discovery of quasinormal oscillations of black holes , many studies have been done on quasinormal modes of various spin fields and a considerable variety of analytical , semi - analytical and numerical methods have been developed to determine them @xcite .
an interesting problem to consider is the study of quasinormal modes of spin @xmath0 fields , called rarita - schwinger fields . in a previous paper shu and
shen @xcite determined the quasinormal modes of this fields in schwarzschild spacetime using the third order wkb method developed by iyer and will @xcite . in this paper
we investigate for the first time the complete time evolution of gravitino perturbations in the schwarzschild background and determine with better accuracy the quasinormal frequencies by fitting time domain data by a superposition of damping exponents , using the prony method .
we also extend previous results by shu and zheng beyond the third order wkb , using the sixth order approach developed by konoplya @xcite .
the results obtained by this two approaches match very well , ensuring the validity of the determined quasinormal frequencies .
we also studied the complete time evolution of scalar , electromagnetic , gravitational and dirac perturbations in the mentioned spacetime , and compute numerically some quasinormal frequencies by the two methods mentioned above .
the structure of this paper is as follows . in section
ii we use the newmann - penrose formalism to obtain the equations for gravitino test perturbations in schwarzschild spacetime , and use separation of variables to obtain decoupled equations for the radial and angular variables involved .
the radial equation is written in the familiar way useful to study the quasinormal spectrum of the perturbations , obtaining an analytical expression for the effective potential related to it .
section iii is devoted to determine the time domain evolution and to compute some quasinormal frequencies of gravitino perturbations , and to compare them with similar calculations for other boson and fermion fields .
the last section contains our conclusions and presents some lines for future research on this subject .
in supergravity theory the gravitational field is coupled to the massless spin 3/2 rarita - schwinger field , that acts as a source of torsion and curvature . if we denote by @xmath1 the spin 3/2 field , then the supergravity fields equations are invariant under supergravity transformations @xmath2 , where @xmath3 is an arbitrary spinor field .
when the rarita - schwinger field vanishes , the supergravity equations becomes einstein s vacuum equations .
therefore , since the torsion and the curvature produced by the rarita - schwinger field depend quadratically on this field , then in the linear approximation about a solution with @xmath4 , the supergravity field equations reduces to the rarita - schwinger field equations together with the vacuum einstein s field equations @xcite . the rarita - schwinger equations for the field @xmath1 in a curved background can be written as @xmath5 where we have defined @xmath6 and @xmath7 defined above , with a multiple principal spinor of the conformal curvature satisfies a decoupled equation @xcite . using the newmann - penrose notation
, we can show that in a frame such that @xmath8 is the only nonvanishing component of the weyl spinor , the rarita - schwinger reduces to only two equations for the components @xmath9 and @xmath10 , that can be solved by separation of variables in all the type d vacuum backgrounds . in terms of the differential operators
@xmath11 , @xmath12 , @xmath13 and @xmath14 that contains the usual directional derivatives of the newmann - penrose formalism , we can put the rarita - schwinger equations in the above mentioned frame as @xmath15 @xmath16 we are interested in the specific case of schwarzschild background , where the metric can be expressed as @xmath17 with @xmath18 and @xmath19 defines the location of the event horizon .
the separable solution of ( [ rseqnp1 ] ) and ( [ rseqnp2 ] ) are given by @xmath20 @xmath21 where @xmath22 is the frequency , @xmath23 is a half integer , and the functions @xmath24 and @xmath25 satisfy the ordinary differential equations @xmath26
@xmath27 where @xmath28 is a separation constant that takes the values @xmath29 with @xmath30 . in the above equations
we have used the operators defined by @xmath31 now introducing the tortoise coordinate @xmath32 defined by @xmath33 and making use of the variable @xmath34 we can put the second equation in ( [ reqs ] ) in the standard form @xmath35 where @xmath36 , @xmath37 and @xmath38 @xmath39 as explained in reference @xcite , it suffices to consider the equation for @xmath40 only , and an entirely similar reduction can be made with the equation for @xmath41 . finally .
it is possible to reduce ( [ eqcasifinal ] ) to a , schrodinger - type equation in the form @xmath42z(r)=0 \label{finaleq}\ ] ] where the effective potential for the rarita - schwinger field is given by @xmath43 with the function @xmath44 defined by @xmath45 , and @xmath46 .
( top ) and @xmath47(bottom ) for @xmath48.,title="fig:",width=415 ] + in figure ( [ potencial ] ) we show the effective potential for gravitino perturbations with two different multipole numbers @xmath49 , for a schwarzschild background with @xmath48 .
as we can see , @xmath50 has the form of a definite positive potential barrier , i.e , it is a well behaved function that goes to zero at spatial infinity and gets a maximum value near the event horizon .
the most direct approach to study the gravitino perturbations is to solve numerically the evolution equation associated with ( [ finaleq ] ) , that is @xmath51 where the function @xmath52 results from the factorization of @xmath53 in ( [ separation2 ] ) as @xmath54 . in order to integrate numerically the equation ( [ evoleq ] ) we use the technique developed by gundlach , price and pulling @xcite , and the result
can be observed as the time - domain profile showed in figure ( [ perfil1 ] ) .
as we can see , the temporal evolution of gravitino perturbation can be divided in three stages .
the first depends on the initial conditions and the point where we observe the profile .
after the initial outburst at the first stage , we observe the exponential damping of the perturbations called quasinormal ringing , that can be split to the superposition of exponentially damping oscillations , represented by a set of complex frequencies , called quasinormal frequencies , whose real parts describe the actual frequency of the oscillation , while the imaginary part is the damping rate . the quasinormal modes are solutions of the wave equation ( [ finaleq ] ) with the specific boundary conditions requiring pure out - going waves at spatial infinity and pure in - coming waves on the event horizon .
thus no waves come from infinity or the event horizon .
as we can see from the figure , the quasinormal ringing stage is followed by the so - called tails at asymptotically late times .
thus , the time evolution of gravitino perturbations in schwarzschild spacetime follows the same stages as well known boson and fermion perturbations .
gravitino perturbations of gravitino modes for schwarzschild black hole at @xmath55 . in the figure the time is measured in units of the horizon radious : @xmath56 .
_ , title="fig:",width=415 ] + to show this , we also integrated numerically the perturbation equations corresponding to scalar , electromagnetic , gravitational and dirac perturbations in this spacetime .
the resulting profiles are observed in figure ( [ perfilspines ] ) .
note that in all of the cases considered , the time evolution is similar , then the gravitino is not the exception . in order to evaluate the quasinormal modes we used two different methods .
the first is a semianalytical method to solve equation ( [ finaleq ] ) with the required boundary conditions , based in a wkb - type approximation , that can give accurate values of the lowest ( that is longer lived ) quasinormal frequencies , and was used in several papers for the determination of quasinormal frequencies in a variety of systems @xcite .
the wkb technique was applied up to first order to finding quasinormal modes for the first time by shutz and will @xcite .
latter this approach was extended to the third order beyond the eikonal approximation by iyer and will @xcite and to the sixth order by konoplya @xcite .
we use in our numerical calculation of quasinormal modes this sixth order wkb expansion .
the sixth order wkb expansion gives a relative error which is about two order less than the third wkb order , and allows us to determine the quasinormal frequencies through the formula @xmath57 where @xmath58 is the value of the potential at its maximum as a function of the tortoise coordinate , and @xmath59 represents the second derivative of the potential with respect to the tortoise coordinate at its peak .
the correction terms @xmath60 depend on the value of the effective potential and its derivatives ( up to the 2i - th order ) in the maximum , see @xcite and references therein . the second method that we used to find
the quasinormal frequencies was the prony method @xcite for fitting the time domain profile data by superposition of damping exponents in the form @xmath61 assuming that the quasinormal ringing stage begins at @xmath62 and ends at @xmath63 , where @xmath64 , then the expession ( [ prony ] ) is satisfied for each value in the time profile data @xmath65 from the above expression , we can determine , as we know @xmath66 , the quasinormal frequencies @xmath67 once we have determined @xmath68 as functions of @xmath69 .
the prony method allows to find the @xmath68 as roots of the polinomial function @xmath70 defined as @xmath71 it is possible to show that the unknown coefficients @xmath72 of the polinomial function @xmath70 satisfy @xmath73 solving the @xmath74 linear equations ( [ prony5 ] ) for @xmath72 we can determine numerically the roots @xmath75 and then the quasinormal frequencies .
it is important to mention the fact that with the prony method we can obtain very accurate results for the quasinormal frequencies , but the practical application of the method is limited because we need to know with precision the duration of the quasinormal ringing epoch . as this stage is not a precisely defined time interval , in practice , it is difficult to determine when the quasinormal ringing begins . therefore , we are able to calculate with high accuracy only two or , sometimes three dominant frequencies . in table
[ frecuencias gravitino ] we show the results for the evaluation of the first four fundamental quasinormal modes for gravitino perturbations in schwarzschild black holes , using the two methods mentioned above . the parameter entering in the calculation is the black hole mass @xmath76 , that we take to be a unit mass .
we also show for comparison the results obtained using a third order wkb method , that were obtained in reference @xcite . with sixth order wkb approximation
we obtain an improve of the results for the quasinormal frequencies , as we can easily see after comparison with the numerical results obtaining using the prony fitting of the time domain data .
the obtained results are showed in figure ( [ qnm ] ) . for a given angular number ,
higher overtones becomes less oscilatory , i.e , the real part of the quasinormal frequencies decreases whereas the imaginary part increases .
in this situation , the quality factor of modes , defined as the ratio @xmath77 , decreases . on the contrary , modes with higher multipole number @xmath49 and the same overtone number have higher frequencies and the damping shows only little increments , with an effective increase of the quality factors .
to test the accuracy of the prony method to finding quasinormal frequencies by fitting time domain data , we also [ cols="^,^,^,^,^ " , ] [ frequencias_gravitacionales ] as we can see , the prony method gives values that are in perfect agreement with previous numerical results existing for scalar , electromagnetic and gravitational perturbations .
also we obtain a improvement of the third order wkb results using the sixth order semianalytic formula .
the dependance of the quality factor with the multipole and overtone number is similar as for the case of gravitino test field perturbations . in figure ( [ qfactor ] ) we show this dependance for gravitino and gravitational perturbations .
for a fixed multipole number , higher overtones shows a decreasing in the quality factor , and the modes are more damping .
in contrast , increasing the multipole number for a given overtone , the quality factor increases due to the increasing of the real oscillating frequency in a more pronounced way that the damping rate . at this point
it is interesting to note that fermion perturbations in the black hole background shows higher quality factors that boson perturbations , as figure ( [ qfactorspin ] ) shows .
as the spin weight of a boson perturbation increases , the quality factor decreases , and the black holes perturbed by this fields becomes more poor oscillators .
then , the lowest quality factors corresponds to gravitational perturbations , and the higher to scalar test perturbations .
the situation for fermion fields is opposite . in this case
the quality factor increases with the increment of the field spin , and as a consequence the highest quality factor belongs to the gravitino perturbations .
then , as a general fact , black holes perturbed by fermion fields are better oscillators than those perturbed by boson fields .
we have studied the evolution of rarita - schwinger field perturbations in a schwarzschild background .
solving numerically the time evolution equation for this perturbations , we find similar time domain profiles as in the case of fields of other spins : the usual three stages in the time evolution dominated at intermediary times by quasinormal ringing .
we determined the quasinormal frequencies by two different approaches , 6th order wkb and time domain integration with prony fitting of the numerical data .
the two methods give close values of qnms for well pronounced potential barriers .
also we apply the prony method to numerically calculated the quasinormal frequencies of scalar , electromagnetic , gravitational and dirac perturbations in the schwarzschild background , with results in perfect agreement with previous numerical calculations by other authors .
the time domain evolution of gravitino perturbations are similar to that corresponding to other fields of different spin .
the behaviour of the quasinormal frequencies with respect to the black holes masses is also similar for all spin perturbations .
however , an interesting difference occur in the oscillating behaviour of black holes perturbed by fields of different spin classes : if the perturbing field has integer spin , then the system becomes a poor oscillator as the spin increases .
the opposite situation occurs in the presence of fermion perturbations , black holes perturbed by higher spin fields are better oscillators . in general ,
the highest quality factor is associated with gravitino perturbations and the smallest with gravitational ones .
there are extensions of this work that are interesting to consider , first the generalization to the case of charged black holes , and second , the determination of the changes of the gravitino quasinormal spectrum for semiclassical solutions , due to the vacuum polarization of quantized fields .
the solution of the above problems will be presented in future reports .
we are grateful to dr .
alexander zhidenko from usp , brazil for providing me with his mathematica code with the implementation of the prony method and valuable information about numerical methods usually employed in the calculation of quasinormal frequencies .
also we would like to acknowledge helpful discussions with elcio abdalla and jeferson de oliveira from department of mathematical - physics at the university of so paulo , where this work was completed .
00 h. p. nollert , _ class .
* 16 * , r159 ( 1999 )
. k. d. kokkotas , d. g. schmidt , _ liv . rev .
* 2 * , 2 ( 1999 ) .
e. berti , v. cardoso and a. o. starinets , _ class .
* 26 * , 163001 ( 2009 ) .
e. berti , v. cardoso and c. m. will , _ phys
. rev . _ * d73 * , 064030 ( 2006 ) .
g. t. horowitz , v. e. hubeny , _ phys .
rev . _ * d62 * , 024027 ( 2000 ) .
a. s. miranda , j. morgan and v. t. zanchin , jhep * 0811 * ( 2008 ) 030 .
s. chandrasekar , _ proc .
a _ * 343 * , 289 ( 1975 ) . s. chandrasekar , _ gravitation _ , ( freeman , san francisco , 1973 ) .
e. w. leaver,_proc .
a _ * 402 * , 285 ( 1985 ) .
e. w. leaver , _ phys .
rev . _ * d45 * , 4713 ( 1992 ) .
e. abdalla , r. a. konoplya , c. molina , _ phys .
rev . _ * d72 * , 084006 ( 2005 ) . c. gundlach , r. h. price , j. pullin , _ phys .
rev . _ * 49 * , 883 ( 1994 ) ; v. cardoso , j. p. s. lemos , _ phys .
rev . _ * d67 * , 084020 ( 2003 ) ; r. a. konoplya , a. zhidenko , _ jhep _ * 0406 * , 037 ( 2004 ) ; r. a. konoplya , _ phys .
rev . _ * d68 * , 124017 ( 2003 ) .
b. wang , c. y. lin , e. abdalla , _ phys .
lett . _ * b481 * , 79 ( 2000 ) ; b. wang e. abdalla , c. molina , _ phys .
rev . _ * d63 * , 084001 ( 2001 ) ; c. molina , d. giugno , e. abdalla , a. saa , _ phys .
_ * d69 * , 104013 ( 2004 ) ; e. abdalla , b cuadros melgar , c. molina , a. pavan , _ nucl .
phys . _ * b752 * , 40 ( 2006 ) . .
r. a. konoplya , _ phys .
rev . _ * d68 * , 024018 ( 2003 ) .
r. a. konoplya , _
* 8 * , 93 ( 2004 ) .
e. berti , v. cardoso , j. a. gonzales and u. sperhake , _ phys .
rev . _ * d75 * , 124017 ( 2007 ) .
a. zhidenko , _ phd thesis _ ,
f. shu and y. shen , _ phys .
_ * b619 * , 340 ( 2005 ) . . . . . .
r. a. konoplya and e. abdalla , _ phys .
_ * d71 * , 084015 ( 2005 ) ; e. abdalla , o. p. f. piedra and j. de oliveira , arxiv:0810.5489;m.i . liu ,
h. i. liu and y. x. gui , _ class .
quantum grav . _ * 25 * , 105001 ( 2008);p . kanti and r. a. konoplya ,
rev . _ * d73 * , 044002 ( 2006 ) ; j. f. chang , j. huang and y. g. shen _ int . j. theor .
_ * 46 * , 2617 ( 2007 ) ; h. t. cho , a. s. cornell , j. doukas and w. naylor , _ phys . rev . _
* d77 * , 016004 ( 2008 ) ; j. f. chang and y. g. shen _ int .
phys . _ * 46 * , 1570 ( 2007 ) ; y. zhang and y. x. gui , _ class .
quantum grav . _ * 23 * , 6141 ( 2006 ) ; r. a. konoplya _ phys .
lett . _ * b550 * , 117 ( 2002 ) ; s. fernandeo and k. arnold , _ gen . rel .
grav . _ * 36 * , 1805 ( 2004 ) ; h. kodama , r. a. konoplya and a. zhidenko , arxiv:0904.2154 ; h. ishihara , m. kimura , r. a. konoplya , k. murata , j. soda and a. zhidenko , _ phys .
rev . _ * d77 * , 084019 ( 2008 ) ; r. a. konoplya and a. zhidenko , _ phys .
_ * b644 * , 186 ( 2007 ) ; r. a. konoplya , arxiv : 0905.1523 ; h. kodama , r. a. konoplya and a. zhidenko , arxiv : 0904.2154 ;
|
we consider the time evolution of massless gravitino perturbations in schwarzschild black holes , and show that as in the case of fields of other values of spin , the evolution comes in three stages , after an initial outburst as a first stage , we observe the damped oscillations characteristic of the quasinormal ringing stage , followed by long time tails . using the sixth order wkb method and prony fitting of time domain data we determine the quasinormal frequencies .
there is a good correspondence between the results obtained by the above two methods , and we obtain a considerable improvement with respect to the previously obtained third order wkb results . we also show that the response of a black hole depends crucially on the spin class of the perturbing field : the quality factor becomes a decreasing function of the spin for boson perturbations , whereas the opposite situation appears for fermion ones .
|
interacting supernovae ( sne ) are stellar explosions that collide with dense circumstellar material ( csm ) produced by the progenitor star . these events are raising critical new questions about the final evolutionary phases of massive stars and the mass - loss episodes that ensue before core collapse ( smith & arnett 2014 ) .
type - iin and ibn sne , in particular , are interacting sne that are characterized spectroscopically by the presence of relatively narrow emission lines of h and he in their spectra ( schlegel 1990 ; filippenko 1997 ; pastorello et al .
2008 ) , which arise from dense csm that becomes illuminated by the shock between the fast moving sn ejecta and slower moving csm ( chevalier & fransson 1994 ) . as such , observations of interacting sne probe the stellar progenitor s pre - sn mass - loss history , providing valuable information on its final evolutionary episodes .
various lines of evidence show that interacting sne require eruptive pre - sn mass loss that is reminiscent of luminous blue variable ( lbv ) stars , like @xmath9 car , although observations indicate a wide range of mass - loss properties ( e.g. , see the review by smith 2014 ) .
the eruptions are often detectable as extragalactic transients , commonly referred to as ` sn impostors ' ( van dyk 2000 ; smith et al . 2011a ; kochanek et al .
2012 ) , which can rival the luminosity of a sn and spectroscopically mimic sne iin .
although the observational distinction between interacting sne and sn impostors is not always clear , a direct link connecting these phenomena has been established for several objects .
the progenitors of sn 2006jc and sn 2011ht were detected undergoing luminous outbursts 12 yr prior to their core - collapse explosions ( foley et al .
2007 ; pastorello et al . 2008 ; fraser et al .
sn 2010mc ( ofek et al .
2013a ) was also detected 12 months before dramatically brightening , although for this case it appears that the precursor event may actually have been the initial phases of the sn caught unusually early ( smith , mauerhan , & prieto 2014 ) .
perhaps the most interesting transient observed to have undergone multiple phases of eruptive mass loss is sn 2009ip .
originally classified as a sn ( maza et al .
2009 ) , this object was actually discovered during an lbv outburst that is , as a sn impostor ( smith et al . 2010 ; foley et al . 2011 ) .
several years of continued activity were followed photometrically and spectroscopically , leading up to sn 2009ip s most extreme outburst in 2012 ( mauerhan et al .
2013 ; prieto et al . 2013
; levesque et al . 2014 ; smith et al . 2013b ; smith , mauerhan , & prieto 2014 ; pastorello et al . 2013 ; fraser et al . 2013b ; margutti et al .
2014 ; graham et al .
the 2012 event was comprised of two main components : the 2012a " phase , marked by an initial peak at @xmath10 mag that lasted for just over 1 month ; and the 2012b " phase , which began with a fast 10-day rise to a second peak of @xmath11 mag , followed by a bumpy decline down to a slowly declining floor in the light curve near @xmath12 mag . based on spectral similarities with known core - collapse sne , mauerhan et al .
( 2013a ) suggested that the relatively faint 2012a phase of sn 2009ip marked the initial stages of a core - collapse sn , while the subsequent 2012b brightening was the result of interaction between this sn and dense csm ejected during the earlier lbv outbursts .
levesque et al .
( 2014 ) also shared the view that the 2012b brightening was the result of interaction between the 2012a outflow and existing csm .
however , several authors have suggested potential alternatives to a core collapse explosion that sn 2009ip s 2012 evolution was possibly the result of one or more nonterminal outbursts ( pastorello et al .
2013 ; fraser et al . 2013b
; margutti et al . 2014 ) , perhaps caused by the pulsational pair - instability mechanism . the motivation for nonterminal scenarios has been based largely on the fact that the total radiated energy of the light curve ( @xmath13 erg ) can be explained by an explosion energy of @xmath14 ergs ( if spherical symmetry is assumed for the csm ) , and also because the late - time data were interpreted as looking different from what is expected for radioactive decay phases .
more recently , smith , mauerhan & prieto ( 2014 ; hereafter smp14 ) showed that the 2012 light curve and spectral evolution are consistent with published models for core - collapse sne , which can be initially faint ( like sn 1987a ) owing to a relatively compact progenitor radius ( i.e. , a blue supergiant , as expected for an lbv , instead of a red supergiant ) , while the late - time characteristics could be explained if the mass of synthesized @xmath15ni was half that of sn 1987a .
smp14 further argued that the radiated energy did not provide an argument against core collapse , since the csm is probably aspherical , leading to an inefficient conversion of kinetic energy into radiation . in line with the spectral modeling results and interpretation of levesque et al .
( 2014 ) , which are consistent with a toroidal distribution of csm , smp14 also proposed a disk - like distribution of csm , but further argued that significant asphericity is required by the fact that broad photospheric lines are still seen at late times , even after strong csm interaction ends .
this requires that a large fraction of the total solid angle of ejecta was able to expand unimpeded by csm .
smp14 also pointed out that the @xmath16100-day persistence of broad photospheric lines requires a large ejecta mass of at least a few m@xmath17 , and is incompatible with a 0.1 m@xmath17 shell , which would become optically thin much more quickly .
a mass of a few m@xmath17 moving at high speeds ( @xmath168000 km s@xmath18 ) directly implies @xmath19 ergs of kinetic energy . in the case of sne
iin , the system geometry is critical if the total radiated energy from csm interaction is to be used to infer the kinetic energy of the sn explosion .
fortunately , spectropolarimetry can yield important clues about the geometry of sne , allowing us to stringently test the hypothesis of aspherical csm . here , we report multi - epoch spectropolarimetry of the 2012 evolution of sn 2009ip , from the end of the initial 2012a phase , through the peak of 2012b , and later into its decline .
the results unambiguously demonstrate that the source of bright continuum emission during the 2012b phase ( i.e. , the csm interaction zone ) was aspherical , consistent with a toroidal or disk - like distribution of csm proposed by smp14 and levesque et al .
furthermore , the results indicate that the initial 2012a phase has a separate polarization component having an axis of symmetry that is distinct from 2012b , which implies that the 2012a event did not create the csm responsible for the 2012b brightening .
diminishing polarization at late times , in addition to similarities in wavelength - dependent polarization across lines for sn 2009ip and other sne , provides additional evidence for a persistent underlying sn photosphere .
altogether , the available evidence supports the hypothesis of a sn explosion for sn 2009ip in 2012 , and argues strongly against nonterminal explosion models for this object ( at least those proposed thus far ) .
spectropolarimetric observations were obtained at the following facilities utilised by the university of arizona : the 6.5 m multiple mirror telescope ( mmt ) on mt .
hopkins during 2012 sep .
21 and 24 ( ut dates are used throughout this paper ) ; the kuiper 61-inch telescope on mt .
lemmon during nov .
11 and 14 ; and the bok 2.3 m telescope on kitt peak during dec . 5 , 6 , and 7 .
observations at these facilities all made use of the same ccd imaging / spectropolarimeter ( spol ; schmidt et al .
the instrument contains a rotatable semiachromatic half - waveplate used to modulate incident polarization and a wollaston prism in the collimated beam to separate the two orthogonally polarized spectra onto a thinned , anti - reflection - coated @xmath20 pixel site ccd .
spol was configured with a 600 line mm@xmath18 grating in first order .
the slit selection was based on the seeing conditions and ranged from @xmath21 to @xmath22 .
this setup resulted in spectral resolutions of @xmath23@xmath24 with useful wavelength coverage in the range @xmath2540007600 .
a standard hoya l38 blocking filter was used to ensure that the first - order spectrum was not contaminated by second - order light for @xmath26 .
the efficiency of the waveplate as a function of wavelength was measured and corrected at all epochs by inserting a fully polarizing nicol prism into the beam above the slit .
the orientation of the slit on the sky was always set to a position angle of @xmath27 , aligned along north - south . a series of four separate exposures that sample 16 orientations of the waveplate yields two independent , background - subtracted measures of each of the normalised linear stokes parameters , @xmath28 and @xmath29 . on some nights
several such polarization sequences of sn 2009ip were obtained and combined , with the weighting of the individual measurements based on photon statistics .
the instrumental polarization of spol and the kuiper , bok and mmt telescopes is extremely small ( @xmath30% ) , as measurements of unpolarized standard stars over the past two decades have consistently shown .
still , to confirm , we observed one or both of the unpolarized standard stars bd+28@xmath314211 and g191b2b ( schmidt et al .
1992b ) during each epoch , confirming that the instrumental polarization has to be less than 0.1% and that there is no low - level residual structure observed in the @xmath28 and @xmath29 spectra .
the linear polarization position angle on the sky ( @xmath32 ) was determined by observing one or more of the interstellar polarization standards hiltner 960 , vi cyg # 12 , or bd+59@xmath31389 ( schmidt et al .
1992b ) during all epochs .
the adopted correction from the instrumental to the standard equatorial frame for @xmath32 for all epochs was determined from the average position angle offset of the polarized standards .
differences between the measured and expected polarization position angles were @xmath33 for all of the standard stars . the sep .
21 mmt observations were impacted by intermittent cloud cover , which degraded the signal - to - noise ratio of these observations .
nonetheless , secure detections of polarization from sn 2009ip were obtained , owing to the dual - beam design of spol , which allows for variations in atmospheric transparency during integrations .
measurements at all other epochs from kuiper and bok were obtained under clear conditions . from arizona latitudes , sn
2009ip was observed through airmass values of 2.052.32 . the shane 3 m reflector at lick observatory equipped with the kast spectrograph ( miller & stone 1993 ) was used on oct . 5 and 14 , and on nov . 6 and 14 .
like spol , kast is a dual - beam spectropolarimeter that utilises a rotatable half - waveplate and wollaston prism .
only the red channel of kast was used for spectropolarimetry .
a gg455 order - blocking filter blocked all second - order light at wavelengths shortward of 9000 .
observations were made with the 300 line mm@xmath18 grating and the 3 slit , yielding a spectral resolution of @xmath1620 .
the orientation of the slit on the sky was always set to a position angle of @xmath27 , aligned along north - south .
exposures of 900 s were obtained at each of four waveplate positions ( @xmath34 , @xmath35 , @xmath36 , and @xmath37 ) . for the oct .
14 , nov . 6 , and nov . 14
observations , three sequences were performed yielding a total on - source exposure time of 3 hr . the observations on oct .
5 were hampered by cloudy weather , and were halted after only one waveplate sequence ( 1 hr total exposure ) . from the latitude of lick observatory ,
sn 2009ip rose to a minimum airmass of 2.47 at transit .
observations typically began at airmass values near 2.5 and ended near 3.2 .
flat - field and wavelength calibration spectra were obtained immediately after each sequence , without moving the telescope . for polarimetric calibrations
, standard stars were selected from the sample of schmidt et al .
( 1992a , b ) .
the unpolarized standard stars bd+32@xmath313739 and hd 57702 were observed to verify the low instrumental polarization of kast .
we constrained the average fractional stokes @xmath28 and @xmath29 values to @xmath38% . by observing bd+32@xmath313739 through a 100% polarizing filter
, we determined that the polarimetric response is so close to 100% that no response correction was necessary , and we obtained the instrumental polarization position - angle curve , which we used to correct the data .
we observed the high - polarization stars bd@xmath3959@xmath31389 and hd 204827 to obtain the zero - point of the polarization position angle on the sky ( @xmath6 ) and to determine the accuracy of polarimetric measurements , which were consistent with published values within 0.05% . since the oct .
5 observations were hampered by intermittent cloud cover , no polarization standard stars could be observed .
thus , for this epoch we assumed negligible instrumental polarization , relying on the fact that the instrumental polarization was below 0.05% for the oct .
14 , nov . 6 , and nov .
14 observations , and we used the position - angle correction curve from oct .
14 , which we have verified to be very stable over time .
we also adopted an offset angle of 5.9@xmath31 to further calibrate the oct .
5 observations ; this is the same offset value measured in the oct . 14 , nov . 6 , and nov .
14 observations , to within 0.5@xmath31 .
all data obtained from the arizona and lick observatories were extracted and calibrated using generic iraf routines and our own idl functions .
spectropolarimetric analysis was also performed in iraf and idl following the methods described by miller et al .
( 1988 ) and implemented by leonard et al .
( 2001 ) .
we also obtained spectropolarimatric data on sn 2009ip from the european southern observatory ( eso ) public archive .
we analysed data from the very large telescope ( vlt ) and focal reducer and low - dispersion spectrograph ( fors2 ) instrument ( appenzeller et al . 1998 ) obtained on 2012 nov .
7 , dec . 6 , and dec .
10 , which were part of the target - of - opportunity programme i d 290.d-5006 .
we used the 300v grism data for nov . 7 and dec .
6 ; these data have spectral resolution of @xmath1613 and wavelength coverage of @xmath1644509500 .
we also analyzed higher - resolution ( @xmath163 ) 1200r grism data for dates nov .
7 and dec . 10 , which covers a wavelength range of 58007300 .
the gg435 order - blocking filter was used for all observations .
flat - field , bias frames , and wavelength - calibration spectra were obtained for both grism settings . a master distortion frame and a waveplate chromatism data file
were also obtained from the archive .
sn 2009ip was observed at airmass values around 1.01.1 for the 300v and 1200r grism data , with respective exposure times of 330 and 470 s per waveplate position for the nov .
observations , and approximately 900 s for the dec . observations .
the position angle on the sky was always kept at 0@xmath31 by the instrument rotator .
standard stars from fossati et al .
( 2007 ) were observed for calibrations .
the high - polarization standard ngc 2024 - 1 was observed on nov .
6 and dec . 7
to calibrate the zero - point of the polarization position angle of the polarimeter on the sky , and the low - polarization standards wd 2039@xmath40202 and wd 2149@xmath39021 were observed on nov . 1 and dec . 7 , respectively , to gauge any instrumental polarization . the instrumental position angle spectrum was obtained directly from the eso public archive . on nov . 1 the observation of wd 2039@xmath40202 yielded linear stokes parameters of @xmath41% , after binning the @xmath0 band , while on dec .
6 , wd 2149@xmath39021 also yielded @xmath41% .
such a small amount of instrumental polarization will not have a substantial impact on the science observations , and we thus make no attempt to remove it . reduction and spectropolarimetric analysis of the vlt / fors2 data was performed using the fors pipeline kit ( version 4.11.13 ) supplied by eso .
specifically , the _ fors_pmos_calib _ and _ fors_pmos_science _ reduction recipes were used , and executed using the _ gasgano _ graphical user interface software .
after all reduction and calibrations were performed , the data were corrected for the redshift of the host galaxy ngc 7259 ( @xmath42 ) .
linear polarization is expressed as the quadratic sum of the @xmath28 and @xmath29 stokes parameters , @xmath43 , and the position angle on the sky is given by @xmath44 , while carefully taking into account the quadrants in the @xmath28@xmath29 plane where the inverse tangent angle is located . since @xmath7 is a positive - definite quantity , it is significantly overestimated in situations where the signal - to - noise is low .
it is typical to express the de - biased " ( or bias - corrected " ) form of @xmath7 as @xmath45 , where @xmath46 and @xmath47 are the uncertainties in the @xmath28 and @xmath29 stokes parameters . if @xmath48 , then we set a 1@xmath49 upper limit on @xmath7 of @xmath50 . in cases
where @xmath51 , @xmath6 is essentially undetermined .
all polarized spectra presented herein are displayed in this manner .
note , however , that at low signal - to - noise ratio @xmath52 is also not a reliable function , as it has a peculiar probability distribution ( miller et al .
thus , for extracting statistically reliable values of polarization within a particular waveband , we have binned the calibrated @xmath28 and @xmath29 stokes spectra separately over the wavelength range of interest before calculating @xmath7 and @xmath6 .
all quoted and tabulated values in this paper were determined in this manner , although spectra are displayed as @xmath52 , so they may exhibit slight offsets from our quoted values .
0.04 for both components , implying low interstellar absorption and , hence , low galactic interstellar polarization ( see 3.1 ) .
na d is seen in weak emission at the redshift of sn 2009ip.,width=316 ] interstellar polarization ( isp ) is commonly a very problematic element of polarimetric analysis , as an accurate probe of the isp from both the milky way and the host galaxy can be very difficult to obtain . fortunately ,
several factors indicate that the isp in the direction of sn 2009ip must be very small , and should not significantly impact our polarimetry results and the physical interpretation that follows .
first , the high - resolution vlt / fors2 flux spectrum on nov . 7 , shown in figure [ fig : na ] , reveals the resolved components of galactic interstellar na i d absorption lines at 5890.1 and 5896.3 . the equivalent width ( ew ) of each component is a very small @xmath53 m , which implies a galactic @xmath54 mag ( munari & zwitter 1997 ) . according to serkowski et al .
( 1975 ) , the value of @xmath55 can be used to derive an upper limit on galactic isp .
the so - called serkowski relation is given by isp @xmath56 , which implies that isp @xmath30% in the direction of sn 2009ip .
second , earlier photometric estimates of the total reddening toward the host galaxy ngc 7259 and sn 2009ip indicate a low value of @xmath57 mag ( schlegel et al .
1998 ; smith et al . 2010 ; foley et al . 2011 ) , which implies that the host galaxy also does not produce substantial isp along the line of sight .
this is also consistent with the lack of na d absorption at the redshift of ngc 7259 ; in fact , na d appears in weak emission , perhaps from a local h ii region or from gas excited by sn radiation , with no indication of additional absorption components .
the low host extinction is not particularly surprising , given sn 2009ip s large radial distance from the center of this nearly face - on spiral galaxy .
if we were to naively apply the serkowski relation to the total measured reddening , we would obtain isp @xmath58% , still a very low value .
however , as shown by leonard et al .
( 2002 ) , the serkowski relation does not necessarily apply to all galaxies outside the milky way , as different dust properties can result in variable efficiencies for isp . nonetheless , the data strongly suggest a low value of total isp , which justifies our decision to make no attempt to remove it from the data , as its effect should be minor and not significantly impact our results .
we revisit the subject of isp at the end of the following section . , and position angle @xmath6 for sn 2009ip during its 2012 outburst , at epochs between sep .
21 and dec .
the three kuiper / spol epochs ( _ red _ ) have been averaged for clarity . the sep .
21 and 24 mmt / spol data ( _ light blue _ ) have been averaged and binned to 300 ; all other epochs are binned to 50 .
note the @xmath59 offset in @xmath6 between the 2012a phase and the peak of 2012b.,width=316 ] -band ( 50505950 ) polarization for sn 2009ip during the 2012a and 2012b phases . for reference , an arbitrarily scaled version of the @xmath0-band light curve
is plotted in the background . _ lower panel _ : evolution in the @xmath28@xmath29 plane .
interstellar polarization is constrained to @xmath600.2% , illustrated by the black dot near the origin .
the dashed lines represent the position angles @xmath61 and @xmath62 measured at the first epoch on sep . 21 and peak polarization on oct .
14 , respectively .
our latest continuum measurement ( 52005500 ) for dec .
6 is marked with a red cross near the origin , consistent with our isp limit .
the approximate point - reflection symmetry between 2012a and the peak of 2012b suggests two separate and roughly orthogonal components of polarization on the sky .
note the temporary shift in @xmath6 associated with the bump in the light curve at @xmath1630 days past peak.,title="fig : " ] -band ( 50505950 ) polarization for sn 2009ip during the 2012a and 2012b phases . for reference , an arbitrarily scaled version of the @xmath0-band light curve
is plotted in the background . _ lower panel _ : evolution in the @xmath28@xmath29 plane .
interstellar polarization is constrained to @xmath600.2% , illustrated by the black dot near the origin .
the dashed lines represent the position angles @xmath61 and @xmath62 measured at the first epoch on sep .
21 and peak polarization on oct . 14 , respectively .
our latest continuum measurement ( 52005500 ) for dec .
6 is marked with a red cross near the origin , consistent with our isp limit .
the approximate point - reflection symmetry between 2012a and the peak of 2012b suggests two separate and roughly orthogonal components of polarization on the sky .
note the temporary shift in @xmath6 associated with the bump in the light curve at @xmath1630 days past peak.,title="fig : " ] figure [ fig : pol_seq ] shows the total flux spectra , @xmath7 , and @xmath6 of sn 2009ip for epochs between sep .
21 and dec . 6 . to gauge the total @xmath0-band polarization @xmath7 and @xmath6 , we first binned @xmath28 and @xmath29 separately in the wavelength range 50505950 .
the extracted values are listed in table 1 .
figure [ fig : pol_lc ] shows the resulting polarization light curve and the temporal evolution in the @xmath28@xmath29 plane .
in addition to @xmath0-band , we also sampled continuum " regions of 52005500 and 61006200 , because they appear mostly devoid of strong line features ; these ranges are close to the respective green " and red " wavelength regions used by hoffman et al .
we caution , however , that there is no location within the spectrum that is completely free of line emission / absorption during all epochs , so the regions we have selected should be regarded as pseudocontinua that avoid the strong lines . [ cols="<,>,^,^,^,^,>",options="header " , ] [ tab : p48 ] @xmath63day is with respect to the day of peak in the @xmath0 band ( jd 2456207.72 ) .
+ @xmath64data for oct . 5 interrupted by poor weather ; required calibration using oct .
14 standard - star observations ; to be used with caution .
+ the sep . 21 and 24 epochs took place near the end of sn 2009ip s 2012a phase , before the sharp rise in brightness that marks the onset of the 2012b phase .
the total flux spectrum during 2012a exhibits the characteristics of sne iin , displaying intermediate - width ( @xmath161000 km s@xmath18 ) emission lines superimposed on a spectrum with broad p - cygni lines ( @xmath168000 km s@xmath18 ) that is reminiscent of more common sne
ii - p ( mauerhan et al . 2013 ) . during this early epoch ,
sn 2009ip is polarized at @xmath160.9% with position angle @xmath2 , and it occupies quadrant iv in the @xmath28@xmath29 plane . by oct . 5 ,
sn 2009ip has entered the 2012b phase and the flux spectrum changes substantially ; the broad emission components have become strongly diluted by the strengthening continuum , which is quite blue , while the intermediate - width emission features have developed broad lorentzian wings , a commonly observed characteristic of sne iin and an indication of high optical depth ( chugai 2001 ; dessart et al .
2009 ; smith et al .
2008 ) , probably resulting from csm interaction . at this time
, @xmath7 has changed only marginally to @xmath160.8% but the position angle has shifted dramatically to @xmath65 , moving into quadrant ii of the @xmath28@xmath29 plane .
interestingly , this dramatic change in position angle indicates separate polarization components for the 2012a and 2012b phases , with distinct axes of symmetry .
the lack of a correspondingly significant change in fractional @xmath7 associated with the large rise in brightness could possibly be explained by the partial cancellation of the polarization vectors from the 2012a and 2012b components , as the latter rises in luminosity . by oct .
11 , the source has risen to peak flux . the spectrum has retained the overall appearance of oct . 5 ,
while the continuum becomes less blue and broad - line p - cygni absorption becomes apparent for he i / na d. our lick / kast and kuiper / spol measurements on oct .
14 show that @xmath7 has risen to a maximum of @xmath161.7% and @xmath6 has reached @xmath66 . at this time , strong deficits in polarization are seen for the intermediate - width emission components of h@xmath67 and , to a lesser extent , h@xmath68 .
this is likely to be the result of dilution from the strong , intrinsically unpolarized emission lines in the outer csm .
meanwhile , a slight increase in @xmath7 is seen at 56005800 , suggesting a potential association with broad he i @xmath695876 or na d p - cygni lines .
figure [ fig : qu_full ] ( top panels ) shows the stokes spectra and the data distribution in the @xmath28@xmath29 plane for the oct .
14 lick epoch , binned to 100 . for reference
, we also show scaled logarithmic flux spectra in the same figure , which has been normalised with a low - order spline ( avoiding the strongest emission features for the spline nodes ) . on oct .
14 , near peak brightness , the entire optical spectrum of sn 2009ip occupies a compact portion of the @xmath28@xmath29 plane , in quadrant ii .
the points form a slightly elongated cluster that shows a continuous trend in wavelength , with the bluest points exhibiting the lowest position angle in the @xmath28@xmath29 plane .
the cluster as a whole exhibits a small but non - negligible spread in position angle , which corresponds to a range of @xmath70 on the sky .
the dilution in polarization associated with intermediate - width h@xmath67 emission is apparent as a protrusion from the cluster of data points which roughly points toward the origin .
note that our binning causes the line to be unresolved , and this affects the strength and position angle of the protrusion ; the unbinned data show the protrusion extending very close to the origin . in total polarization @xmath7 , we see broad , shallow variations that appear to be associated with broad absorption components of h@xmath67 and he i / na d blueward of 5876 .
it is interesting to note that the scaled flux spectrum included in figure [ fig : qu_full ] shows that the broad p - cygni emission features from fast ejecta , first seen during 2012a , are still present throughout the peak of 2012b ; this has important physical consequences for the energetics of sn 2009ip ( see 4 ) . by nov . 67 , roughly 30 days past peak ,
the continuum has faded substantially and broad p - cygni lines have once again become prominent in the flux spectrum .
he i / na d absorption has deepened , and the intermediate - width component of h@xmath67 has developed an absorption component on top of the broad underlying feature , perhaps part of its own p - cygni profile .
interestingly , this epoch coincides with the appearance of a temporary albeit substantial bump in the declining light curve .
7 vlt / fors2 spectropolarimetry data are shown in detail in the lower panels of figure [ fig : qu_full ] . at this time
, the @xmath0-band polarization has dropped to @xmath160.7% and the position angle has shifted substantially to @xmath1650@xmath31 .
the enhancement of polarization near he i / na i absorption feature has become more pronounced and sharp ; this feature is also apparent in lick data from nov .
6 , shown in figure [ fig : pol_seq ] .
meanwhile , the sharp dip in polarization produced by the dilution from unpolarized intermediate - width balmer lines is no longer apparent .
we examine the polarization enhancements of two specific line features in figure [ fig : qu_line1 ] , which show the stokes parameters plotted as a function of velocity , with respect to the he i @xmath695876 and h@xmath67 rest wavelengths . on nov .
7 , the enhanced polarization of he i / na d is substantially blueshifted to a velocity of 13,000 km s@xmath18 ( if it is associated with he ) and exhibits a velocity width of @xmath161500 km s@xmath18 . the peak is also blueshifted by @xmath161500 km s@xmath18 with respect to the flux minimum of the absorption line . the enhanced he i / na d feature forms a linear - shaped loop that roughly traces the average polarization axis of the system for that date
. there also appears to be an enhancement in fractional polarization for the absorption component blueward of the intermediate - width h@xmath67 emission feature ( at @xmath71 km s@xmath18 ) .
the origin of this absorption feature is unclear .
it could be an absorption component of a p - cygni profile for the intermediate - width h@xmath67 component , or it could be the result of blueshifted absorption for he i @xmath696678 . if it is from h@xmath67 p - cygni , then this enhanced polarization feature does not appear blueshifted as in the case of the broad he i / na d absorption . in the @xmath28@xmath29 plane
, the track across this potential h@xmath67/he i feature appears to be roughly consistent with the overall axis of symmetry .
there also appears to be some indication of polarization enhancement for blend of absorption components seen at the higher velocities of @xmath72 km s@xmath18 and larger , which are probably from h@xmath67 .
overall , the broad emission component of h@xmath67 appears to make a loop in the @xmath28@xmath29 plane . averaging all of our measurements from kuiper / spol on nov .
12 and 14 , we detect @xmath160.7% polarization in the @xmath0 band , similar to the value of the preceding vlt / fors2 epoch on nov .
7 . the lick / kast data solely from nov .
14 show a value of polarization similar to the nov.1214 combined average of the kuiper / spol data .
the polarized spectra are not shown in figure [ fig : pol_seq ] , to avoid cluttering the figure , although their broadband measurements are presented in figure [ fig : pol_lc ] .
interestingly , although the overall degree of polarization has not changed substantially between nov . 6 and 14 , the position angle has shifted again back toward the axis of symmetry of the 2012b peak .
it thus appears that the bump in the light curve on nov .
67 is associated with another component of polarization , perhaps related the 2012a luminosity source .
a few weeks later , on dec .
57 , we detect weak polarization in an average of three separate measurements with bok / spol , obtaining @xmath73 and only an upper limit for the continuum of @xmath74% . the last vlt / fors2 measurement takes place on dec .
6 , for which we obtain @xmath75% and @xmath76% . for the december
epochs the position angle has changed significantly with the diminishing polarization , moving into quadrant iii with @xmath77 .
the flux spectrum during this late epoch shows that the continuum has become redder , while the broad p - cygni features are still present .
he i / na d absorption has become a double - dipped profile .
figure [ fig : qu_line2 ] shows that by dec .
6 the enhanced polarization of this feature has shifted redward , and is probably associated with the bluer component of the double - dipped profile . a complicated blend of multiple absorption and perhaps emission features appears blueward of h@xmath67 at this time , and also exhibits evidence for enhanced polarization , roughly along the average axis of symmetry .
examining more closely the temporal evolution of sn 2009ip in the @xmath28@xmath29 plane , shown in figure [ fig : pol_lc ] ( lower panel ) , the range of possible isp values constrained by the value of @xmath55 ( see 3.1 ) are illustrated by the radius of the black spot centered on the origin .
it seems plausible that the path traced by @xmath28 and @xmath29 during dec .
is the result of a migration back to the point of isp , somewhere near the origin .
indeed , our latest measurement of the continuum polarization yielded @xmath74% from bok / spol for dec .
57 , and @xmath78% from vlt / fors2 on dec .
6 ( @xmath79% , @xmath80 ) , which is consistent with the isp limits we have estimated . an alternative , and less likely possibility , is that @xmath28 and @xmath29 are destined to drift back to the location in quadrant iv occupied by our earliest mmt / spol measurements in september ( i.e. , that the 2012a phase is intrinsically unpolarized , and instead reflects the location of isp ) . but such a large value of isp ( @xmath811% ) seems implausible given the constraints on isp indicated by the low values of @xmath55 and ew(na d ) ; the structure in @xmath7 for the september epochs also does not follow a serkowski law , which is inconsistent with the 2012a polarization being the result of isp .
furthermore , it would be a rather remarkable coincidence for the position angle of isp to be orthogonal to the polarization axis of sn 2009ip at peak .
therefore , the evidence is strong that our earliest polarization detections from mmt / spol in september reveal true polarization intrinsic to the source during the 2012a phase , and that the isp is within @xmath58% of the origin , and likely @xmath82% , as indicated by our latest continuum measurement .
as one can see , any value of isp within this allowable range will affect our results only slightly , and will not change the fact that the 2012a and 2012b phases appear to have two separate components of polarization that will remain roughly orthogonal between 2012a and the peak of 2012b , no matter where the isp lies within the range illustrated by figure [ fig : pol_lc ] ( lower panel ) .
the most natural explanation for linear polarization of a sn is thomson scattering by free electrons , either from the sn photosphere or within the zone of intense csm interaction , which includes the forward shock , the reverse shock , and the cold dense shell ( cds ) in between . for most sne iin , including sn 2009ip , evidence of electron scattering is also indicated by the broad lorentzian wings typically seen at the bases of the intermediate - width emission - line profiles ( chugai et al .
2001 ; smith et al . 2008 ; dessart et al . 2009 ) .
but in the case of scattered photons emerging from a spherically symmetric distribution of material , the sum of all electric vectors will cancel out and yield zero net polarization .
thus , the detection of net polarization from any sn directly implies aspherical structure , either intrinsic or imposed from partial obscuration of the polarized surface by intervening material ( see wang & wheeler 2008 for a review on sn polarization ) .
core - collapse sne of type ii - p have been observed to be weak sources of polarization while in their plateau phases , which implies that the outer layer of ejecta is roughly spherical in some cases .
however , near the end of their plateaus , sne ii - p can exhibit brief but significant increases in polarization ( leonard et al . 2000 , 2006 ; chornock et al .
2010 ) , because as the ejecta cool and recombine the photosphere recedes from the spherical outer layers of h - rich material into more aspherical he - rich and metal - rich layers deeper inside the ejecta .
thus , to first order , sne ii - p can be characterized by a single component of photospheric polarization that scales inversely with the decreasing luminosity .
sne iin , on the other hand , appear to exhibit a high degree of polarization while in their brightest phases , at least for the few objects that have been studied in sufficient detail ( e.g. , leonard et al .
2001 ; hoffman et al .
this implies that csm interaction luminosity must be a major source of the net polarization from sne iin . in this case , since both the sn photosphere and the csm interaction zone are potential sources of polarized photons , changes in the relative luminosities of these components can lead to variations in net polarization , even if the fractional polarization and position angles of each component remain constant with time .
below , we propose this to be the case for sn 2009ip .
sn 2009ip s photospheric component is made evident by its spectropolarimetric similarities with the type iib sn 1993j , shown in figure [ fig:93j ] . near peak brightness , both sne exhibit a comparable degree of polarization across the optical band , with sn 2009ip being stronger in the continuum , and they both display remarkably similar wavelength - dependent variations across their polarization spectra , particularly evident at wavelengths of 52007000 . in the case of sn 1993j , these variations are associated with the broad p - cygni lines of h@xmath67 and he i@xmath39na d , and the same is probably true for sn 2009ip .
the variations are produced by the depolarizing effect of line scattering ( trammell et al .
1993 ; hflich 1995 ; hflich et al .
1996 ; tran et al .
this is not to be confused with the intrinsically unpolarized recombination emission from the outer optically thin csm that we see in sn 2009ip , which dilutes the polarized continuum and results in relatively narrow absorption " features of polarization for the h@xmath67 and h@xmath68 lines .
rather , true depolarization is important for the broad emission lines that develop in the ionised ejecta above the electron - scattering photosphere of the sn .
the persistence of these lines in sn 2009ip for @xmath16100 days implies a massive amount of high - velocity material ( a few m@xmath17 ) , consistent with a sn outflow ( see smp14 ) .
csm is also a potential contributor to the polarization properties of sn 1993j ( hflich et al .
1996 ; tran et al .
1997 ) , via electron scattering within a sn / csm interaction zone and/or the scattering of photons by dust .
after all , the continuum polarization of sn 1993j being entirely due to an aspherical ejecta photosphere is somewhat difficult to reconcile with the evolution of a highly spherical radio remnant revealed by vlbi ( bietenholz , bartel , & rupen 2003 ) , although this structure might only trace the morphology of the outermost h - rich layers , not the more asymmetric he - rich layers potentially responsible for the polarized continuum .
still , the source did exhibit evidence for pre - existing csm at early times , in the form of narrow emission - line features exhibiting small velocity widths of @xmath16170 km s@xmath18 , indicative of an ambient stellar wind ( benetti et al .
1994 ) . however , these features were less pronounced than those of sn 2009ip and lasted for a shorter amount of time , indicating a much less dense csm .
sn 1993j s x - ray properties were also indicative of csm interaction ( leising et al 1994 ) , while its radio evolution indicated a changing pre - sn mass - loss rate in the range @xmath83@xmath84 yr@xmath18 ( van dyk et al .
1994 ) , which , although roughly 100 times weaker than the pre - sn mass - loss rate inferred for sn 2009ip ( ofek et al .
2013b ) , could generate enough csm to influence the object s polarization properties . therefore , like sn 2009ip , sn 1993j could have two components of polarization photospheric and csm interaction . in the case of sn 2009ip
, however , it appears rather obvious that csm interaction is the dominant source of polarization near photometric peak .
the opposite could be true for sn 1993j : its polarization could be dominated by an aspherical explosion and only moderately influenced by the csm component . in this regard ,
a noteworthy difference between the two sne is that near peak brightness sn 1993j exhibits more variation than sn 2009ip in @xmath6 as a function of wavelength , suggesting that sn 2009ip has a more highly ordered axis of symmetry in 2012b resulting from its luminous and highly aspherical csm interaction zone . by 5060 days after peak , however , both sne exhibit remarkable similarity in @xmath7 and @xmath6 as a function of wavelength .
it appears that as csm interaction fades after the peak of 2012b and the broad p - cygni lines return by nov .
6 , the polarization characteristics of sn 2009ip begin to look more like a sn photosphere again .
it is thus interesting that the source makes a pronounced shift in position angle on nov .
6 , associated with the obvious bump in the light curve ( see figure [ fig : pol_lc ] ) , which could represent the temporary brightening of the photospheric component of polarization that dominated the 2012a phase . taken at face value , sn 2009ip s degree of peak polarization for the 2012a and 2012b phases ( from @xmath1% to @xmath85% ) indicates respective axial ratios of @xmath600.85 and @xmath600.7 for the aspherical source ( @xmath8615% and @xmath8630% asphericity ) , if we assume that polarization translates directly to the apparent geometry via comparison with the oblate electron - scattering atmosphere models of hlfich ( 1995 ) .
however , the true physical axial ratios could be substantially more aspherical than these limits , for two key reasons : ( 1 ) inclination angle , and ( 2 ) partial cancellation between the 2012a and 2012b polarization vector components . if both the 2012a peak and the late decline of the 2012b phase ( from the nov .
67 bump onward ) are dominated by the sn photosphere component , as appears to be the case , then partial cancellation could be particularly important for the 2012b peak .
indeed , the @xmath59 shift in position angle between the earliest 2012a observation and the peak of the 2012b phase implies two separate components of polarization that are roughly orthogonal to one another . since orthogonal polarization vectors cancel , the degree of intrinsic polarization for the 2012b component must be higher than it appears ( i.e. , even more aspherical ) , because the orthogonality of the separate 2012a vector component , if still present during 2012b , will partially cancel the 2012b component . we can estimate the true intrinsic polarization of the 2012b peak component as follows . if we rotate the @xmath28@xmath29 values from figure [ fig : pol_lc ] such that they are aligned with the axis of symmetry at maximum polarization ( @xmath87 ) , which roughly traces the average axis of symmetry for the entirety of 2012b , then the majority of polarization during the 2012b phase is shifted onto the positive @xmath28 axis , while the orthogonal 2012a polarization gets shifted onto the negative @xmath28 axis .
figure [ fig : rsp ] shows the evolution of the resulting rotated stokes parameters .
the @xmath88 curve represents variations along the main axis of symmetry , while @xmath89 traces deviations from the average axis of symmetry .
the result implies that if the 2012a component maintains its orthogonal contribution to the total electric vector during 2012b , then the separation between the 2012a and 2012b components will approximately define the true degree of intrinsic polarization for the 2012b peak , which , therefore , could be as high as @xmath162.6% . in this case ,
comparison with the models of hlfich ( 1991 ) would imply a more highly aspherical axial ratio of @xmath900.6 for the 2012b component ( @xmath8640%50% asphericity ) . considering the additional effect of a @xmath91 inclination angle , hypothetically , it is possible that the intrinsic geometry of the 2012b component is a highly flattened one .
below we discuss a plausible geometric scenario for sn 2009ip s 2012 evolution , facilitated by figures [ fig : cartoon1 ] and [ fig : cartoon2 ] . ) .
the plot is meant to illustrate the full possible degree of intrinsic polarization for the 2012b component ( @xmath162.6% ) , after considering the effect of the partial cancellation , roughly orthogonal 2012a vector component ( see 4.1.2).,width=316 ] for the 2012b component of polarization the situation appears relatively clear . since csm interaction dominates the luminosity during this phase , and since the associated polarization source exhibits a rather steady and well - defined axis of symmetry , the net polarization must reflect the physical structure of the dense csm , which in this case could represent a toroidal geometry ( a disk or ring ) . toroidal geometry for the interacting csm
has already been suggested for sn 2009ip by levesque et al .
( 2014 ) , based on spectral modeling and analysis of the balmer emission lines .
those authors found that the h@xmath67 emission during the peak of 2012b requires a total radiating area of @xmath1620,000 au@xmath92 , while the balmer decrement requires a high csm density of @xmath93 @xmath94 , for case - b recombination ( note that such a high density would naturally explain the absence of forbidden emission lines in sn 2009ip s spectrum ) .
the combination of high density and large area implies a flattened geometry ( i.e. , a disk / ring ) , possibly as thin as 10 au .
alternatively , a lower density csm of @xmath95 @xmath94 in a limb - brightened spherical shell configuration was also proposed by levesque et al .
( 2014 ) as an additional possibility , although less favourably because it requires that the optical depth of h@xmath68 be lower than for h@xmath67 , which would be unusual .
independently , smp14 also favoured disk - like geometry for the csm based on energetic arguments and on the persistence of broad p - cygni lines even after the phase of intense csm interaction had declined , which implies that a large fraction of the ejecta did not decelerate and , thus , must not have participated in strong csm interaction ( toroidal geometry would allow a large fraction of the ejected mass to bypass strong csm interaction in directions orthogonal to the csm plane ) .
our spectropolarimetric results are consistent with toroidal geometry for the csm around sn 2009ip , and contradict models that invoke spherical distributions of csm . in this case , the axial ratio of the polarized emission source at the peak of 2012b ( constrained to @xmath96 , and probably @xmath600.6 ; see 4.1.2 ) is consistent with toroidal geometry tilted at some intermediate inclination angle with respect to our line of sight .
the spectropolarimetric evolution is _ inconsistent _ with some physical scenarios previously proposed as possibilities for sn 2009ip s 2012 evolution . since it is the csm geometry that determines the polarization properties at the peak of the interaction phase , then the earlier eruption that produced such csm should reflect the same geometry .
thus , the orthogonal position angle shift we observe between 2012a and the peak of 2012b is not compatible with the interpretation that these respective phases represent the launching and collision of successive shells of csm ( e.g. , from pair - instability eruptions ) , a possibility suggested by other authors ( pastorello et al .
2013 ; margutti et al .
2014 ) , since such a scenario should result in similar axes of symmetry for the two phases , not orthogonal geometries .
rather , the orthogonality between the 2012a and 2012b events implies a physically distinct origin for each of the associated outflows .
in addition , the large shift in polarization and position angle between the 2012b peak and the subsequent bump in the declining light curve on nov .
67 also suggests that the bump is the result of a polarized luminosity component that is separate from that of the 2012b peak for example , the sn photospheric component brightening again as the csm interaction component fades away .
this is inconsistent with the interpretation that the bump in the light curve is the result of a fluctuating central source that variably illuminates the csm , such as activity from a central surviving star , as proposed by martin et al .
( 2013 ) .
toroidal csm geometry would not be unique to sn 2009ip ; it has been proposed before for other sne iin , specifically in the cases of sn 1988z ( chugai & danziger 1994 , their model b ) , sn 1998s ( leonard et al . 2000 ) , and sn 1997eg ( hoffman et al .
this type of geometry is also unsurprising in the context of sn 2009ip s stellar progenitor , since rings and tori are common phenomena around lbvs , b[e ] supergiants , and other evolved massive stars ( smith , bally , & walawender 2007 ; smith et al .
not at all clear , however , is the connection of such geometry to the erratic behaviour of sn 2009ip over the last decade .
it is possible that dense equatorial csm is the result of binary - influenced mass loss , as has been suggested to explain the toroidal csm of b[e ] supergiants .
a connection between such objects and sn 2009ip was speculated upon by clark et al .
( 2013 ) in the case of lha 115-s 18 , a b[e ] supergiant exhibiting erratic long - term variability that is reminiscent of lbvs .
it is thus interesting to note that the radial dimensions of b[e ] disks ( @xmath97@xmath98 au ; zickgraf et al .
1986 ) are comparable to the size estimates for sn 2009ip s toroidal csm ( smith et al . 2013 ; smp14 )
. represents the angle subtended by the csm from the point of view of the explosion.,width=316 ] and [ fig : qu_line2]).,width=316 ] the origin of the polarization for the 2012a phase is less clear . since the spectrum of sn 2009ip on sep .
2124 resembles that of more common core - collapse sne , the simplest explanation might be electron scattering within the photosphere of the early sn , which has yet to impact the dense csm lying farther out . in this case
, the axial ratio of @xmath99 and the orthogonal geometric axis with respect to the peak of the csm interaction phase would suggest a prolate / bipolar shape for the 2012a outflow .
interestingly , bipolar geometry has been suggested previously for other cases of highly polarized core - collapse sne ( e.g. , see wang et al .
2001 ) . alternatively ,
net polarization from the 2012a phase could also arise if the near side of the soon - to - be - shocked toroidal csm is partially blocking the waist of the sn photosphere ( figure [ fig : cartoon1 ] , inset ) , which could result in a polarization axis that is orthogonal to 2012b if the polarized poles of the photosphere remain relatively unobscured .
finally , scattering of the sn photons by dust in the outer csm could provide another source of polarized photons .
after all , infrared data obtained during the 2012a phase show strong evidence for circumstellar dust at a radius of @xmath16120220 au ( smith et al .
we note that the strength of sn 2009ip s blue continuum detected at the same time indicates that the central source is probably not completely enshrouded by the dusty component , which is consistent with the toroidal geometry inferred for the csm .
however , if dust scattering from the dense csm were an important source of polarized flux during 2012a , we would expect the geometric axis during this phase to reflect that of 2012b rather than being orthogonal to it .
thus , electron scattering within the sn photosphere appears to be the most plausible explanation as the dominant source of polarized photons during 2012a , with net polarization either being the result of bipolar geometry for the photosphere or partial blocking of its waist by the toroidal csm .
if strong csm interaction is confined to a toroidal geometry , then large amounts of high - velocity sn ejecta could expand relatively unimpeded .
this is already indicated by the persistence of the broad p - cygni lines ( smp14 ) , but could also explain some of the higher - order effects we see in the spectropolarimetric data .
for example , the sharp blueshifted enhancement in polarization associated with the he i / na d feature seen on nov .
67 ( figure [ fig : qu_line1 ] ) is an indication of high - velocity material .
this feature has a blueshifted velocity of 13,000 km s@xmath18 , while the position angle across the feature approximately follows along the main axis of symmetry of this epoch in the @xmath28@xmath29 plane .
the fact that the peak of the enhanced polarization feature is not coincident with the flux minimum of the absorption line implies that the enhancement can not simply be the result of absorption of unpolarized or less polarized flux ( which would increase the fractional polarization at that wavelength ) .
rather , the net blueshift probably results from an intervening layer of fast he - rich ( or na - rich ) material that lies near the outer edge of the ejecta and is _ partially _ absorbing the inner polarized continuum . from the point of view of the observer , such partial blocking can increase the net polarization over that range of velocities by interfering with the cancellation of orthogonal polarization vectors distributed across the polarized source ( e.g. , also see kasen et al .
the high - velocity he / na d - rich material was probably launched during the sn in 2012a phase , while the csm responsible for the 2012b phase was likely to have been created by the lbv progenitor eruptions during the years prior ( e.g. , see graham et al .
2014 , their figure 7 ) .
a toroidal or disk - like distribution of csm around sn 2009ip has important consequences for the kinetic energy of the explosion that is inferred from the total radiated energy measured during the 2012b csm - interaction phase .
taking the 2012a explosion time of @xmath100 days , suggested by smp14 , and the fastest ejecta velocity of 13,000 km s@xmath18 indicated by the blue edge of the h@xmath67 p - cygni line during 2012a , implies a radial distance of @xmath16350 au from the point of explosion to the inner edge of the dense csm ; this is consistent with the limits on the minimum radius of the dust component inferred from infrared measurements during 2012a ( @xmath86120 au ; smith et al .
2013b ) , and with the 300 au radial estimates by levesque et al .
the latter authors showed that the csm need only be @xmath1610 au thick to explain the h@xmath67 luminosity in the toroidal scenario .
for a 10 au toroidal scale height , the 300350 au distance from the explosion would imply a very small subtended angle of @xmath101 for the csm from the point of view of the explosion . such a configuration inclined at an angle of 3545@xmath31 with respect to our line of sight ( from edge - on )
would appear as an ellipse on the sky having an axial ratio that is consistent with the geometric constraints placed by the value of peak polarization during 2012b ( see 4.1.2 ) .
for such a small subtended angle of csm , the fraction of the total explosion solid angle ( @xmath102/4@xmath103 ) intercepted by csm will be only 23% ( @xmath6010% conservatively ) .
this minor amount of interacting area will result in an inefficient conversion of the total kinetic energy of the sn into radiation .
thus , the @xmath104 ergs of kinetic energy inferred from the radiated energy of the explosion by fraser et al .
( 2013 ) and margutti et al .
( 2014 ) , both of whom assume a spherical csm configuration in their calculations , must be a significant underestimate , by 1 dex or more .
a kinetic energy of @xmath19 ergs , as suggested by smp14 , seems more likely , given the demand for aspherical csm by spectropolarimetry .
one must also keep in mind that the shape of the explosion is an influential factor as well . a bipolar explosion , which we have shown is a plausible geometry for the 2012a outflow , will intersect even less area of the toroidal csm .
if the material is clumpy , then the effective intersecting area could be even lower than we have estimated , implying an even lower efficiency conversion of kinetic energy into radiation .
we have presented multi - epoch spectropolarimetry of the 2012 outburst of sn 2009ip , covering the 2012a and 2012b phases .
since the available data imply a very low amount of interstellar absorption and isp @xmath58% , we have not attempted to remove isp in our analysis .
this could potentially lead to some small systematic errors in our results , although we expect any such changes will be minor and will not significantly influence our interpretation .
the degree of polarization at the peak of 2012b implies an approximate axial ratio of @xmath600.7 ( probably @xmath600.6 ) for the polarized csm , which is consistent with the toroidal distribution of material suggested previously ( smp14 ; levesque et al .
2014 ) . in the future ,
modeling of the polarization properties of shocked toroidal csm could reveal how well the comparison to the oblate electron - scattering atmosphere models of hfilch ( 1991 ) can characterise the axial ratio of such geometry .
details aside , the spectropolarimetric results demand substantial asphericity for the source of the 2012b luminosity and are thus inconsistent with models for sn 2009ip s 2012 explosion that invoke a spherical distribution of csm ( fraser et al .
2013 ; margutti et al . 2014 ) .
for the 2012a phase , the detection of significant polarization at a position angle that is nearly orthogonal to 2012b is consistent with a bipolar geometry for the sn outflow or partial obscuration of a potentially more spherical sn photosphere by the toroidal csm . in any case , the orthogonality is inconsistent with the launching and collision of successive shells of csm ( pastorello et al . 2012 ) generated @xmath10540 days apart ( margutti et al .
2014 ) , since such a scenario should produce similar axes of symmetry for the 2012a and 2012b phases , not orthogonal geometries .
furthermore , the geometric parameters constrained by spectropolarimetry imply that only a small fraction of the sn ejecta ( @xmath6010% ) participated in strong interaction with the toroidal csm . therefore , earlier estimates of a @xmath104 ergs explosion that are based on the total radiated energy , which assume spherical symmetry for the csm ( fraser et al .
2013 ; margutti et al . 2014 ) , are likely to be substantially underestimated by an order of magnitude .
the strong similarities between sn 2009ip and sn 1993j , from the time near their peak through their decline , provide another strong line of evidence for the existence of an underlying sn photosphere from high - velocity ejecta ( 50008000 km s@xmath18 ) that persists throughout sn 2009ip s 2012 evolution ( @xmath16100 days ) .
such a long - lasting optically thick component indicates a mass of at least a few m@xmath17 for the fast outer component of the ejecta , and the associated speeds imply a kinetic energy of @xmath19 ergs for the ejecta , as suggested by smp14 .
after peak polarization , higher - order structure in the @xmath28@xmath29 plane from he - i / na d indicates the presence of fast metal - rich ejecta that have overrun the dense csm , consistent with a high - velocity flow that bypassed intense csm interaction .
meanwhile , the weakening of polarization with declining continuum flux at late times suggests that the diminishing intensity of electron scattering is responsible for the decline in polarization .
the fact that the flux and polarization drop more rapidly for sn 2009ip relative to other sne iin , such as sn 1997eg ( hoffman et al .
2008 ) or sn
2010jl ( grant williams , 2014 , private communication ) , also suggests that the csm around sn 2009ip has a compact configuration , compared to these other explosions .
the lines of evidence presented here thus favour a scenario first proposed by mauerhan et al .
( 2013 ) , in which the luminosity of the 2012a phase of sn 2009ip was the result of a @xmath19 ergs explosion ( i.e. , a core - collapse sn ) that subsequently plowed into an aspherical ( probably toroidal ) distribution of csm during 2012b , creating the jump in luminosity and the strong polarization of the source .
nonterminal scenarios are difficult to support in light of the current body of evidence , and we would have to invent a process capable of generating ejecta having @xmath19 ergs of kinetic energy without destroying the star . still , the best evidence of progenitor death is a vanishing of the star at late times .
however , csm interaction can persist for decades or more , and generate enough luminosity to make progenitor disappearance a difficult observable to confirm .
sn 1961v is a particularly controversial example ( filippenko et al . 1995 ; chu et al . 2004 ; smith et al . 2011b ; kochanek et al .
2011 ; van dyk et al .
moreover , it is very plausible that the progenitor of sn 2009ip s 2012 explosion left a massive companion at the explosion site , not only because massive stars are rarely solitary ( sana et al .
2012 ) , but because toroidal or disk - like geometry for the csm is suggestive of binary influence . as discussed by smith & arnett ( 2014 ) , repeated brief pre - sn eruptions could result from binary interaction if the primary suddenly increases its radius during late nuclear burning stages , thereby triggering binary interaction and mass ejection . finally , we note that strong net polarization consistently observed from sne iin , as a class , implies that csm interaction is a dominanting influence on their polarization properties , and that an aspherical distribution of csm is a commonplace feature of the stellar progenitors of these explosions .
we thank the referee for their review of this manuscript .
we thank the staffs at lick , the mmt , and steward observatories for their excellent assistance .
we thank s. bradley cenko for assisting with the lick observations . some observations reported here
were obtained at the mmt observatory , a joint facility of the university of arizona and the smithsonian institution .
this research was also based , in part , on observations made with eso telescopes at the la silla paranal observatory under programme i d 290d-5006 .
we thank hien tran for supplying polarimetric data on sn 1993j for our comparison .
thanks dan kasen at uc berkeley for insightful discussion , and we also thank leah n. huk at u. denver for helpful commentary .
this research was supported by nsf grants ast-1210599 ( ua ) , ast-1211916 ( uc berkeley ) , ast-1210311 ( sdsu ) , and ast-1210372 ( u. denver ) .
the supernova research of a.v.f.s group at u.c .
berkeley is also supported by gary & cynthia bengier , the richard & rhoda goldman fund , the christopher r. redlich fund , and the tabasgo foundation .
fossati , l. , bagnulo , s. , mason , e. , & landi deglinnocenti , e. 2007 , the future of photometric , spectrophotometric and polarimetric standardization , 364 , 503 fraser , m. , magee , m. , kotak , r. , et al .
2013a , apjl , 779 , l8 smith n. , chornock r. , li w. , et al .
2008 , apj , 686 , 467 smith n. , miller a. , li w. , et al . 2010 , aj , 139 , 1451 smith n. , et al . 2011a , mnras , 415 , 773 smith n. , mauerhan , j.c . ,
kasliwal , m. , & burgasser , a. 2013 , mnras , 434 , 2721
|
we present spectropolarimetry of sn 2009ip throughout the evolution of its 2012 explosion . during the 2012a phase ,
when the spectrum exhibits broad p - cygni lines , we measure a @xmath0-band polarization of @xmath1% at a position angle of @xmath2 , indicating substantial asphericity for the 2012a outflow . near the subsequent peak of the 2012b phase ,
when the spectrum shows signs of intense interaction with circumstellar material ( csm ) , we measure @xmath3% and @xmath4 , indicating a separate component of polarization during 2012b , which exhibits a higher degree of asphericity than 2012a and an orthogonal axis of symmetry on the sky .
around 30 days past peak , coincident with a substantial bump in the declining light curve , we measure @xmath5% and another significant shift in @xmath6 . at this point ,
broad photospheric lines have again become prominent and exhibit significant variations in @xmath7 relative to the continuum , particularly he i / na i d. by 60 days past peak the continuum polarization has dropped below 0.2% , probably declining toward a low value of interstellar polarization .
the results are consistent with a scenario in which a prolate ( possibly bipolar ) explosion launched during the 2012a phase impacts an oblate ( toroidal ) distribution of csm in 2012b .
previous calculations that assumed spherical symmetry for the csm have substantially underestimated the required explosion energy , since only a small fraction of the sn ejecta appears to have participated in strong csm interaction . an ejecta kinetic energy of at least @xmath8 ergs is difficult to avoid , supporting the interpretation that the 2012 outburst of sn 2009ip was the result of a core - collapse explosion .
[ firstpage ] circumstellar matter stars : evolution stars : winds , outflows supernovae : general supernovae : individual ( sn 2009ip )
|
since the experiments revealed the @xmath0-wave symmetry of the order parameter in cuprate superconductors @xcite , the physics of low - energy quasiparticle excitations in disordered two - dimensional ( 2d ) @xmath0-wave superconductors has been a subject of ongoing intensive research @xcite .
the characteristic feature of the @xmath2-wave pairing state is the existence of four nodal points where the order parameter vanishes . in the vicinity of the gap nodes there
exist low - lying dirac - type quasiparticle excitations .
an understanding of the disorder effect induced by randomly - distributed impurities on these low - energy states is essential for the elucidation of the thermodynamic and transport properties of disordered @xmath0-wave superconductors . during the years
a number of theoretical approaches , such as the self - consistent approximation schemes [ 39 ] , non - perturbative methods [ 1017 ] , and numerical studies [ 18 - 22 ] , have been developed to calculate the quasiparticle density of states ( dos ) in the presence of disorder .
unfortunately , many of these theories contradict each other .
the dos at zero energy was variously predicted to be finite [ 310 ] , vanishing [ 1114,18,21 ] , and divergent [ 15,20 ] .
recently , it has been made clear that [ 16,17,19,22 ] much of this controversy roots in the fact that the @xmath0-wave superconductor is fundamentally sensitive to the details of disorder , as well as to certain symmetries of the normal state hamiltonian . in order to clarify the physics of the various asymptotic results for the dos , yashenkin et al .
@xcite analyzed the diffusive modes in disordered @xmath0-wave superconductors and calculated the weak - localization correction to the dos with the diagrammatic technique .
it is well known that the weak localization in electron systems is a manifestation of the quantum interference ( qi ) effect , which stems from the existence of the goldstone modes ( cooperon and diffuson ) @xcite .
as pointed out by altland and zirnbauer @xcite , the goldstone modes in a superconductor have different features from those in a normal metal , for the local particle - hole symmetry of the superconducting state gives rise to a combination of impurity- and andreev - scattering processes . as a result
, every cooperon or diffuson mode in the retarded - advanced ( ra ) channel entails a corresponding mode in the retarded - retarded ( rr ) or advanced - advanced ( aa ) channel . in the combined limit of the unitarity and nested fermi surface ( the un limit ) ,
each of these @xmath3-mode cooperon and diffuson has a @xmath4-mode counterpart due to the global particle - hole symmetry @xcite .
for disordered @xmath0-wave superconductors , it was found @xcite that in generic situations the quasiparticle dos is subject to a logarithmic weak - localization correction due to the 0-mode cooperon , and the existence of diffusive @xmath4-modes can give rise to a finite , or even divergent zero - energy dos upon approaching the un limit .
the weak - localization calculations carried out by yashenkin et al . appear to capture the physical origin of the discrepancies between predicted low - energy quasiparticle dos .
furthermore , the quasiparticle weak - localization effect was also suggested to have important influences on transport properties such as the electrical @xcite , spin @xcite , thermal @xcite , and optical @xcite conductivities of @xmath0-wave superconductors .
while the qi effects have been widely investigated for disordered normal metals @xcite , a corresponding theory for random dirac fermions in superconducting cuprates is far from well established , and thus highly deserves further development .
this paper presents an intensive study of the qi effect on the quasiparticle dos in 2d @xmath0-wave superconductors with dilute nonmagnetic impurities both near the born and near the unitary limits .
these two limiting cases are considered to be closely related with the disorder effects in cuprate superconductors .
it is reasonable that the disorder due to defects off the copper - oxygen plane may be treated in the born approximation and that defects in the plane may be in the unitary limit @xcite .
albeit sharing certain aspects with ref .
@xcite , we further develop the weak - localization theory in @xmath0-wave superconductors , and obtain some new results in this paper .
first , the expressions of the goldstone modes for quasiparticle diffusion are derived in details both near the born and near the unitary limits .
second , we calculate the additional contributions to the dos from those lowest - order self - energy diagrams containing non - singular ladders , which were not taking into account previously . these diagrams are shown to give rise to various renormalization factors for the dos correction in the born and unitary limits .
third , by taking into account a new nontrivial self - energy diagram with the @xmath4-mode diffuson , we show that the qi correction to the dos becomes a @xmath1-function of the energy upon approaching the un limit .
this result can be used to account for the resonant peak found by the previous numerical studies @xcite .
the structure of this paper is as follows . in sec .
ii , the commonly used self - consistent @xmath5-matrix approximation ( sctma ) is described for a weakly - disordered @xmath2-wave superconductor . using the sctma
, we derive in sec .
iii the expressions of the 0-mode and @xmath4-mode cooperons and diffusons .
the qi correction to the quasiparticle dos is calculated in sec .
iv , and the conclusions are summarized in sec .
v. the appendix provides some mathematical formulas , which are useful for our derivations .
let us consider a most extensively studied model for a 2d @xmath2-wave superconductor . in the tight - binding approximation ,
the normal - state dispersion of a square lattice is given by @xmath6 where @xmath7 is the lattice constant , @xmath8 is the nearest - neighbor hopping integral , and @xmath9 is the chemical potential .
the nested fermi surface corresponds to the half - filling case ( @xmath10 ) .
the order parameter of the @xmath2-wave pairing state can be expressed by @xmath11 .
the four gap nodes are given by @xmath12 with @xmath13 , which satisfy @xmath14 .
in the vicinity of these nodes , the quasiparticle spectrum can be linearized as @xmath15 , where @xmath16 , @xmath17 , and @xmath18 is the momentum measured from the node @xmath19 .
a direct calculation yields @xmath20 . in the presence of randomly - distributed nonmagnetic impurities , the time - reversal and spin - rotational symmetries
remain preserved .
thus the system belongs to symmetry class ci in the classification of ref .
the impurity potential is assumed to be point - like , meaning that the intra- and inter - node scatterings are described by a single potential strength @xmath21 . for a low impurity concentration @xmath22
, the quasiparticle self - energy in the sctma @xcite can be expressed in the nambu spinor representation as @xmath23 for @xmath24 . here
@xmath25 is the mass renormalization factor , @xmath26 is the impurity - induced relaxation rate , @xmath27 is a dimensionless parameter , @xmath28 represents the decrease of the chemical potential induced by the impurity scatterings , and @xmath29 and @xmath30 @xmath31 stand for the @xmath32 unity and pauli matrices , respectively .
a use of dyson s equation immediately yields the impurity - averaged one - particle green s functions as @xmath33\tau_0+\delta_{\bm k}\tau_1+\xi_{\bm k}\tau_3}{[(1-\lambda)\epsilon\pm i\gamma]^2-\epsilon^2_{\bm k}},\ ] ] where the shift of chemical potential has been absorbed in @xmath9 .
the zero - energy density of states is calculated as @xmath34 , where @xmath35 and @xmath36 with @xmath37 .
the parameters @xmath26 , @xmath25 , and @xmath27 can be evaluated consistently via the @xmath5-matrix equation @xmath38 with @xmath39 .
using eq .
( 2.2 ) , we can show that @xmath40\tau_0 + \left(v^{-1}-u^{-1}\right)\tau_3,\ ] ] for @xmath24 , where @xmath41 is the effective impurity potential given by @xmath42 . a substitution of eqs .
( 2.1 ) and ( 2.4 ) into eq . ( 2.3 ) leads to @xmath43 , @xmath44 , and @xmath45 .
the born limit corresponds to @xmath46 , yielding @xmath47 and the unitary limit corresponds to @xmath48 , meaning that @xmath49 throughout this paper , we mainly consider the cases near either the born or the unitary limit .
it is worthy to point out that the values of the impurity potential @xmath21 driving the system into these two limits are dependent on the band structure .
since the qi effect is related to the diffusive modes , we first derive the expressions of the 0-mode and @xmath4-mode cooperons and diffusons for a disordered @xmath0-wave superconductor .
the @xmath3-mode cooperon and diffuson exist both in ra and in rr channels due to the local particle - hole symmetry @xmath50 .
the ladder diagrams for the cooperon are given by fig .
1(b ) in ref . @xcite .
the equation for @xmath3-mode cooperon can be expressed as @xmath51 where the two - particle irreducible vertex @xmath52 and the integral kernel @xmath53 are defined in the rr and ra channels as @xmath54 and @xmath55 equation ( 3.1 ) can be also expressed as @xmath56 where @xmath57 with @xmath58 . from eqs .
( 3.1)(3.5 ) , it follows that one can make a decomposition of @xmath59 for @xmath60 , and @xmath61 . substituting eq .
( 2.1 ) into eq .
( 3.2 ) , we can easily obtain all the nonvanishing components of @xmath62 as @xmath63,\ ] ] @xmath64 @xmath65 and @xmath66 it then follows that the dominant component of @xmath52 near the born limit ( @xmath67 ) is @xmath68 , while that near the unitary limit ( @xmath69 ) is @xmath70 .
now let us calculate @xmath71 and @xmath72 . for small values of @xmath73 , @xmath74 , and @xmath75 , we have @xmath76 and @xmath77 substituting eqs . ( 3.10 ) and ( 3.11 ) into eq . ( 3.3 ) , and using eqs .
( a1)(a4 ) in the appendix , we can show that all the nonvanishing components of @xmath71 are given by @xmath78 @xmath79,\ ] ] and @xmath80.\ ] ] the diagonal components of @xmath72 can be easily calculated by a substitution of eqs .
( 3.6 ) , ( 3.7 ) , and ( 3.12)(3.14 ) into eq . ( 3.5 ) , yielding @xmath81,\end{aligned}\ ] ] @xmath82,\end{aligned}\ ] ] @xmath83,\ ] ] and @xmath84.\end{aligned}\ ] ] we note that all the non - diagonal components of @xmath53 are vanishing , and those of @xmath52 are negligible near the born or unitary limit
. then eq .
( 3.1 ) or ( 3.4 ) indicates that only the diagonal components of @xmath85 may be singular for these two limits . as a result
, the 0-mode cooperon can be expressed as @xmath86 . as will be shown below
, all @xmath87 are of diffusive poles .
obviously , only the diagonal components of @xmath88 and @xmath62 are needed for the calculation of 0-mode cooperon , and thus eq . ( 3.4 ) becomes equivalent to the following group of equations : @xmath89 @xmath90 @xmath91 @xmath92 where the arguments ( @xmath93 , @xmath74 , and @xmath75 ) of @xmath94 , @xmath95 , and @xmath96 have been omitted .
near the born limit ( @xmath46 ) , the singular terms in eqs .
( 3.19)(3.22 ) satisfy the following relations : @xmath97 @xmath98 @xmath99 @xmath100 the only solution of which is given by @xmath101 substituting eqs .
( 3.7 ) , ( 3.15)(3.18 ) , and ( 3.27 ) into eq . ( 3.22 ) , and using eq . ( 2.5 ) , we can show that the terms of order @xmath102 cancel exactly out , leading to @xmath103 where @xmath104 is the quasiparticle diffusion coefficient . combining eq .
( 3.27 ) with eq . ( 3.28 ) , and noting that the @xmath3-mode diffuson has the same expression as that of the @xmath3-mode cooperon due to the time - reversal symmetry , we obtain @xmath105 which is in agreement with that appearing in ref .
@xcite ( aside from a disputed pre - factor ) .
similarly , we can show that eq .
( 3.29 ) is also valid near the unitary limit ( @xmath106 ) .
the above evaluations indicate that any small deviations from either limit do not make the goldstone 0-modes gapped .
however , in the intermediate region far from these two limits , the non - diagonal components of @xmath62 can not be neglected , and thus @xmath85 may contain some singular non - diagonal components . in the un limit , there exist the additional @xmath4-mode cooperon and diffuson due to the global particle - hole symmetry @xcite @xmath107 , with @xmath108 the nesting vector . any small deviations from this combined limit can be shown to make the goldstone @xmath4-modes gapped .
the ladder diagrams for the @xmath4-mode cooperon can be obtained from those of the @xmath3-mode cooperon by replacing @xmath93 by @xmath109 .
the equation for the @xmath4-mode cooperon is given by @xmath110 where @xmath111 with @xmath112 in order to calculate @xmath113 and @xmath114 , one needs to exploit the relations @xmath115 and @xmath116 .
for a nearly - nested fermi surface @xmath117 , we have @xmath118 substituting eqs . ( 3.11 ) and ( 3.33 ) into eq .
( 3.32 ) , and using eqs .
( a1)(a4 ) , we can show that all the nonvanishing components of @xmath113 are given by @xmath119 @xmath120 @xmath121 and @xmath122 a substitution of eqs .
( 3.6)(3.9 ) and ( 3.34)(3.37 ) into eq .
( 3.31 ) yields the diagonal components of @xmath123 as @xmath124,\end{aligned}\ ] ] @xmath125,\end{aligned}\ ] ] @xmath126,\ ] ] and @xmath127.\end{aligned}\ ] ] like the @xmath3-mode cooperon , the @xmath4-mode cooperon near the born or unitary limit can be also expressed as @xmath128 , with all @xmath129 assumed to be of diffusive poles .
therefore , eqs .
( 3.19)(3.22 ) are also suitable for the @xmath4-mode cooperon .
near the unitary limit ( @xmath130 ) , the singular terms in eqs .
( 3.19)(3.22 ) ( for the @xmath4-mode cooperon ) satisfy the following relations : @xmath131 @xmath132 @xmath133 @xmath134 the solution of which is shown to be @xmath135 substituting eqs . ( 3.6 ) , ( 3.38)(3.41 ) and ( 3.46 ) into eq .
( 3.19 ) , and using eq .
( 2.6 ) , we can show that @xmath136 with @xmath137 .
the present expression of @xmath1 supports the result estimated by yashenkin et al .
@xcite . combining eq .
( 3.46 ) with eq . ( 3.47 ) , and noting that the @xmath4-mode diffuson has also the same expression as that of the @xmath4-mode cooperon due to the time - reversal symmetry , we obtain @xmath138 near the born limit ( @xmath46 ) , one can easily show that @xmath139 , indicating that the diffusive @xmath4-modes exist only near the un limit .
contrary to the diffusive 0-modes , the goldstone @xmath4-modes are gapped by any small deviations from the un limit measured by @xmath1 . for the situations
far from the un limit , the contributions of diffusive @xmath4-modes to the qi effect are completely suppressed due to the large gap .
the qi correction to the quasiparticle dos can be calculated via @xmath140 where @xmath141 represents the lowest - order correction to the one - particle green s function due to the diffusive modes .
the lowest - order self - energy diagrams with 0-mode cooperon are depicted in fig . 1 . the contribution of fig .
1(a ) has been calculated in ref .
figure 1(b ) contains the non - singular ladders , its contribution to the dos can be shown to be the same order of magnitude as that of fig .
according to eq .
( 4.1 ) , the contribution of fig .
1(a ) to the dos reads @xmath142 by means of eq .
( a5 ) , it can be rewritten as @xmath143_{ii},\ ] ] with @xmath144 . the contribution of fig .
1(b ) can be similarly evaluated , and the total contribution of figs .
1(a ) and 1(b ) can be expressed by @xmath145_{ii},\ ] ] with @xmath146 here the matrix @xmath147 satisfies @xmath148 with @xmath149 . substituting eqs .
( 2.1 ) and ( 2.2 ) into eq . ( 4.4 ) , and using eqs .
( a1)(a4 ) , we can show that @xmath150 where @xmath151 equation ( 4.5 ) leads to @xmath152 , and hence @xmath153 , indicating that the contribution of fig .
1(b ) just renormalizes the pre - factor of @xmath154 . a substitution of eqs .
( 2.2 ) and ( 4.5 ) into eq . ( 4.3 ) yields @xmath155 substituting eqs . ( 3.29 ) and ( 4.7 ) into eq . (
4.2 ) , and noting that the upper cutoff of @xmath156 can be set to be @xmath157 with @xmath158 the mean - free path , we obtain @xmath159 where @xmath160 .
the lowest - order self - energy diagrams with the @xmath3-mode diffuson are given by figs .
2(a ) and 2(b ) , and the vertex correction to the impurity - scattering @xmath5-matrix is shown by fig .
2(c ) . unlike the single - vertex - dressed diagram used in ref .
@xcite , the present figs .
2(a ) and 2(b ) include the additional diagrams with _ both _ vertices dressed by the @xmath4-mode diffuson .
similar vertex correction has been considered in the theory of disordered interacting - electron systems @xcite .
the vertex - corrected retarded @xmath5-matrix can be expressed by @xmath161 where @xmath162 , and @xmath163 are the indices of the nambu space , and the vertex function @xmath164 is given by @xmath165 in order to calculate the vertex function , we exploit the equation for 0-mode diffuson in the rr - channel @xmath166 yielding @xmath167 a substitution of eq .
( 4.12 ) into eq . ( 4.10 ) leads to @xmath168 making use of eqs .
( 3.6 ) and ( 3.7 ) , we get @xmath169 substituting eqs .
( 3.29 ) and ( 4.14 ) into eq . ( 4.13 ) , we obtain the expression of the vertex function as ( for @xmath170 and @xmath171 ) @xmath172 which is suitable both near the born and near the unitary limits . substituting eqs .
( 2.1 ) and ( 4.15 ) into eq .
( 4.9 ) , we can easily show that @xmath173 indicating that both figs .
2(a ) and 2(b ) have vanishing contributions to the dos near the born or unitary limit .
near the un limit , besides the diffusive 0-modes , the goldstone @xmath4-modes also contribute to the qi effect .
the leading self - energy diagrams with the @xmath4-mode cooperon can be obtained from those in fig . 1 by a replacement of @xmath174 . similarly with the case of 0-mode cooperon , the contribution of @xmath4-mode cooperon to the dos can be evaluated via @xmath175_{ii},\ ] ] where @xmath176 with @xmath177 for the unitary limit .
expression ( 4.18 ) can be readily evaluated by means of the global particle - hole symmetry , yielding @xmath178 substituting eqs .
( 3.48 ) and ( 4.19 ) into eq . ( 4.17 ) , we obtain @xmath179 comparing eq .
( 4.8 ) with eq . ( 4.20 ) , one finds that the contribution to dos of the @xmath4-mode cooperon has an equal magnitude but an opposite sign to that of the 0-mode cooperon in the un limit ( @xmath180 ) .
the leading self - energy diagrams containing the @xmath4-mode diffuson are generated by replacing @xmath93 by @xmath109 in figs .
2(a ) and 2(b ) .
obviously , eqs .
( 4.9 ) and ( 4.13 ) are also suitable for the @xmath4-mode diffuson . by exploiting eq .
( 3.48 ) , we can readily show that @xmath181 and @xmath182 which are valid near the un limit .
the contribution of the @xmath4-mode diffuson to the dos is given by @xmath183.\ ] ] substituting eqs .
( 2.2 ) and ( 4.22 ) into eq . ( 4.23 ) , and using eqs .
( 2.6 ) and ( a1)(a4 ) , one can easily show that @xmath184 expression ( 4.24 ) is quite different from the logarithmic behavior obtained in ref .
@xcite , due to the additional contributions of the self - energy diagrams with both the impurity - scattering vertices dressed by the @xmath4-mode cooperon . for generic fermi surfaces ,
the qi correction to the dos results only from the contribution of 0-mode cooperon , i.e. , @xmath185 where @xmath186 is given by eq .
therefore , the qi effect gives rise to a logarithmic suppression to the quasiparticle dos , as predicted by yashenkin et al .
equation ( 4.25 ) is suitable near the born or unitary limit . in the intermediate region
far from these two limits , however , the 0-mode cooperon may contain some singular non - diagonal components , and thus a refined theory is necessary for the generic situations .
in addition , the additional contributions of the self - energy diagrams with non - singular ladders yield various renormalization factors in these two limits . near the un limit
, the qi correction to the dos is the sum of @xmath187 , @xmath188 , and @xmath189 , so that @xmath190.\ ] ] the above equation indicates that the quasiparticle dos near the un limit can be subject to a positive correction due to the contributions of the goldstone @xmath4-modes . upon approaching the un limit ( @xmath1 is small enough ) , the contributions from the 0-mode and @xmath4-mode cooperons cancel out each other , so that the dos correction is given only by the contribution of the @xmath4-mode diffuson , i.e. , @xmath191 this result can be used to account for the sharp peak found in the numerical studies of the dos @xcite .
we have extensively investigated the qi effect on the quasiparticle dos in a @xmath2-wave superconductor with dilute nonmagnetic impurities . as in the study of disordered normal metals @xcite ,
the understanding of the diffusive modes is essential for the investigation of the qi effect in a superconductor . through detailed derivations ,
we have obtained the expressions for the diffusive modes both near the born and near the unitary limits .
it is demonstrated that these goldstone modes may contain non - diagonal components in the intermediate region far from these two limits .
therefore , a refined weak - localization theory is necessary for the generic situations . for generic fermi surfaces , the qi effect results only from the 0-mode cooperon , yielding a logarithmic suppression to the quasiparticle dos near the born or unitary limit .
we show that those non - trivial self - energy diagrams containing non - singular ladders give rise to various renormalization factors of the dos correction for the born and unitary limits . near the un ( unitary and nesting ) limit ,
the qi effect comes not only from the contribution of the 0-mode cooperon , but also from those of the @xmath4-mode cooperon and diffuson .
it is found that the self - energy diagrams with both impurity - scattering vertices corrected by the @xmath4-mode diffuson have an important contribution to the dos , which is proportional to @xmath192 . as a result ,
the quasiparticle dos is subject to a positive correction induced by the diffusive @xmath4-modes . upon approaching the un limit ( @xmath193 ) , the contributions of the 0-mode and @xmath4-mode cooperons cancel out each other ( even including the renormalization contributions induced by the non - singular ladders ) ,
so that the dos correction becomes a @xmath1-function of the energy .
this result can be used to account for the resonant peak found in the previous numerical studies @xcite .
a similar sharp peak of electronic dos has been found numerically in a disordered 2d tight - binding model for the normal state @xcite .
the appearance of this sharp peak can be also explained by the contribution of fig .
2(a ) with the @xmath4-mode diffuson ( fig .
2(b ) has a vanishing contribution for the normal state ) @xcite . like in the case of disordered interacting - electron systems @xcite ,
all the leading polarization diagrams responsible for the qi effect on the quasiparticle conductitity can be generated in the conserving approximation from the lowest - order self - energy diagrams shown in figs . 1 and 2 .
for the normal state , it has been shown that the contributions of those polarization diagrams with @xmath4-mode diffuson lead to an antilocalization correction to the conductivity in the un limit @xcite .
how the qi processes related with the diffusive @xmath4-modes affect the transport properties , such as electrical and spin conductivities , in a @xmath0-wave superconductor is another interesting problem , and will be studied in our future work .
this work was supported by the national natural science foundation of china under grants no .
10274008 and no .
10174011 , and the jiangsu - province natural science foundation of china under grants no .
bk2002050 and bk2001002 .
dyx would like to acknowledge the support of grant no .
g19980614 for state key programs for basic research of china . in this appendix
, we present some mathematical formulas , which are useful for the evaluations in the previous sections . by the approach used in ref .
@xcite , we can show that @xmath194 @xmath195 @xmath196 and @xmath197 as an example , we shall prove eq .
( a1 ) . using the dirac - type quasiparticle spectrum , and noting that there exist four gap nodes
, we have @xmath198 by means of the transformations of @xmath199 and @xmath200 , the above equation can be changed as @xmath201 ^ 4}\nonumber\\ & = & \frac{1}{2\pi^2v_fv_g\gamma^2}\int^{x_0}_{0}dx\frac{x^2}{(1+x)^4}\int^{2\pi}_{0}d\theta \cos^2\theta\sin^2\theta,\nonumber\end{aligned}\ ] ] where @xmath202 with @xmath203 .
for the weak - disorder case ( @xmath26 is small enough ) under considered , we can set @xmath204 .
thus , a completion of the integrals over @xmath205 and @xmath206 in the above equation immediately yields eq .
( a1 ) .
another useful formula is given by @xmath207 where @xmath208 , and @xmath209 with @xmath210 and @xmath211 being any linear superimpositions of @xmath30 ( @xmath212=0,1,2,3 ) . in order to prove eq .
( a5 ) , we assum that @xmath213 and @xmath214
. then we have @xmath215 and @xmath216 from eq .
( a7 ) , it follows that @xmath217 a combination of eq .
( a6 ) with eq .
( a8 ) immediately leads to eq .
|
the quantum interference effect on the quasiparticle density of states ( dos ) is studied with the diagrammatic technique in two - dimensional @xmath0-wave superconductors with dilute nonmagnetic impurities both near the born and near the unitary limits .
we derive in details the expressions of the goldstone modes ( cooperon and diffuson ) for quasiparticle diffusion .
the dos for generic fermi surfaces is shown to be subject to a quantum interference correction of logarithmic suppression , but with various renormalization factors for the born and unitary limits . upon approaching the combined limit of unitarity and nested fermi surface ,
the dos correction is found to become a @xmath1-function of the energy , which can be used to account for the resonant peak found by the numerical studies .
pacs numbers : 74.25.jb , 71.20.-b , 73.20.fz
|
the static and dynamic behaviour of sliding charge density waves ( cdws ) is perhaps the most well studied example of a class of problems involving the transport of an elastic medium through a disordered background .
the cdw , which behaves like an elastic medium , is pinned by impurities distributed randomly throughout the material . as the magnitude of an externally applied electric field
is varied , a depinning transition is seen , from a stationary phase at weak fields , to a moving phase at strong fields where the cdw slides through the material . in the vicinity of this depinning transition ,
the dynamics of the cdw are correlated over long distances , with characteristic correlation lengths diverging at the threshold field .
it has been shown@xcite that the behaviour near the threshold field can be studied as a critical phenomenon associated with a second order phase transition.@xcite extensive numerical simulations@xcite on a number of classical models for cdws have helped in understanding the critical properties in the vicinity of the depinning transition . in particular ,
critical exponents describing the scaling of physical quantities can be determined for various spatial dimensions of the cdw .
recently , it has also proved possible to obtain analytically the behaviour in the moving phase above threshold , through a renormalization group treatment.@xcite the results , obtained within an @xmath2-expansion for @xmath3 spatial dimensions , agree very well with the numerical simulations . while the critical behaviour above threshold is fairly well understood , it is more difficult to carry out an analytical treatment below threshold .
the reason why the analysis is simpler in the moving phase is that , for models in which the elastic interactions within the cdw are strictly convex , and no dislocations ( `` phase slips '' ) are allowed , the system approaches a unique steady - state configuration at long times ( up to time translations).@xcite a perturbation expansion around some relatively simple configuration is thus more likely to succeed . in the stationary phase ,
on the other hand , the impurities can pin the cdw in any of several different states ; the particular state that the cdw reaches depends on the past history of the system .
any analytical treatment must include this history dependence , instead of being the simple search for a single attractor , as is sufficient above threshold .
the history dependence of the cdw below threshold is most easily seen in the polarizability .
if the driving force @xmath4 ( produced by the external electric field ) is raised monotonically towards the depinning threshold @xmath5 , the response @xmath6 to infinitesimal increases in the force becomes larger and larger , and diverges at @xmath5 as @xmath7 , with an exponent @xmath8 .
also , at any value of @xmath4 , the response of the cdw to increasing the force further is dominated by localized regions of activity , or `` avalanches '' , whose characteristic size can be used to define a diverging correlation length @xmath9 .
if instead of this monotonic approach , the force is _ lowered _ from just below threshold , the polarizability @xmath10 does not diverge at @xmath5 ( although , depending on the specific model used for the cdw , it may have a cusp singularity ) .
it is thus necessary to distinguish between the upwards polarizability , @xmath6 , and the downwards polarizability @xmath10 . in this paper
, we show how it is possible to obtain analytically the critical behaviour for the ` natural ' monotonic approach to threshold , through an expansion around the mean field solution with the proper history dependence .
our results are in good agreement with recent numerical simulations , as well as new data that we present here .
we find that the correlation length exponent is given by @xmath11 for a @xmath12 dimensional system , while the polarizability exponent is @xmath13 these exponents can in fact be obtained for a highly simplified cdw model@xcite that can be solved exactly .
however , the dynamic exponent @xmath14 _ can not _ be obtained from this model : we show that @xmath14 is equal to its value above threshold @xmath15 in @xmath3 dimensions . later in this paper , we discuss how the model of ref
. corresponds to a pathological limit , and why , despite this , it obtains @xmath16 and @xmath8 correctly .
the solution that we obtain is similar in form to the solution above threshold.@xcite however , there are complications with two - sided scaling .
this is because , as was seen above threshold , there is an extra relevant operator at the fixed point of the renormalization group , which affects only the statics and not the dynamics .
this operator has very different consequences in the pinned and moving phases of the cdw , resulting in different correlation length exponents on the two sides of the phase transition .
the absence of two - sided scaling is unlike the behaviour for phase transitions in equilibrium systems , and conventional arguments about the properties of scaling functions must be used with caution here . in `` phase - only '' models for cdws
, the cdw is described by a periodic modulation in the density of electrons in the material ; the phase of this modulation is assumed to be the only dynamical variable of importance and is assumed to be a continuous function of position .
in numerical simulations , the system is discretized to a lattice , with a continuous phase variable at each site .
although simulations on such continuous variable models have been performed@xcite , it is much more efficient to simplify the model , with the phase of the cdw at any lattice site restricted to being an integer multiple of @xmath17 away from a ( site dependent ) preferred phase , and time being incremented in discrete steps.@xcite the resulting automaton models , which differ in details of how these approximations are made , yield far more precise results . within the numerical uncertainties ,
much of the critical behaviour is the same for both classes of models . in particular , the correlation length exponent @xmath16 and ( with larger numerical uncertainties )
the polarizability exponent @xmath8 appear to be the same .
an important exception to this is the downward polarizability , @xmath18 : as already mentioned , the nature ( or even the existence ) of a cusp singularity in @xmath18 _ does _ depend on the details of the model used , and is in fact even different in different continuous variable models .
however , this difference is not seen in the monotonic approach , which is what we shall be interested in here . since the approach to threshold causes the system to become increasingly unstable , with larger and larger avalanches and a divergent polarizability , it is perhaps not surprising that differences between automaton and continuous dynamics do not affect the critical behaviour .
an important exception to this rule is the distribution for the _ number _ of avalanches . for continuous dynamics ,
an infinitesimal increase in @xmath4 triggers avalanches in which the change in the phase at any point is bounded above by @xmath17 ( corresponding to unity in the automaton models ) .
most of the avalanche area advances by almost @xmath17 . in large avalanches , it is highly probable that a subsequent small increase in @xmath4 will `` retrigger '' avalanches in the same region , as the original unstable point is likely to be nearly unstable at the completion of the first avalanche .
avalanches of usually decreasing size will be retriggered until the region is more stable . in the automaton models , owing to the discretization of the cdw phases
, several `` retriggered '' avalanches are grouped together into a single large avalanche .
this difference between the two classes of models is one that involves the behaviour at _ short times _ ; this does not alter long - wavelength low - frequency characteristics that determine @xmath16 and @xmath14 ( nor @xmath8 , which involves a spatial average over the entire system ) .
however , it may affect avalanche distributions , which do depend on how avalanches are grouped .
in this paper we present numerical data on the distribution of avalanches for the automaton models , and examine the differences with results for continuous variable models .
the automaton models for cdws are closely related to some of the sandpile models that have been proposed and studied recently.@xcite we find that the avalanche distribution that we obtain here for the cdw automaton models agrees very well with the numerical results on sandpiles at the critical point.@xcite the dynamic exponent , @xmath14 , that relates the characteristic duration of an avalanche to its size , is also found to be the same numerically for both classes of models : @xmath19 the _ approach _ to threshold for sandpile models has been treated through numerics and scaling arguments by tang and bak@xcite ; our results differ in certain respects from theirs , which we discuss in detail later in this paper . the connection between sandpiles and cdw models can be exploited to obtain the dynamic exponent @xmath14 for two dimensional systems .
majumdar and dhar@xcite have shown that this exponent is equal to @xmath20 for sandpile models in @xmath21 ; this is in fair agreement with the numerical results for cdws.@xcite they are also able to relate the two exponents that enter the avalanche distribution to each other , thus leaving only one of the two to be determined numerically . in this paper
we show that it is in fact possible to obtain the same exponent identity from completely different scaling arguments on cdws , if we assume that the total rate of avalanche generation ( which is dominated by _
small _ avalanches ) is not singular as threshold is approached .
( we have verified this assumption numerically for the automaton models . ) the rest of this paper is organized in the following manner .
section ii carries out the analytical treatment of the behaviour below threshold , and obtains the critical exponents .
section iii examines the numerical results , which agree quite well with the analytical predictions .
section iv compares the distribution of avalanches for cdw continuous dynamics , cdw automata , and sandpile models , as well as exploring other connections to sandpile models .
section v discusses the relationship of this work to earlier results , and its possible relevance to other physical systems .
the model that we use in this section is the fukuyama - lee - rice model,@xcite which assumes that the distortions of the cdw are continuous ( _ i.e. _ that there are no dislocations ) .
the dynamics can then be expressed in terms of a phase variable , @xmath22 , which measures the distortions with respect to an ideal undistorted cdw .
( the position @xmath23 is a @xmath12-dimensional vector . ) assuming that the dynamics are strongly overdamped , and are given by a simple relaxation of an energy functional @xmath24 , the equation of motion is @xmath25 in eq.([flr ] ) there are three terms on the right hand side : a simple elastic force that arises from an elastic energy @xmath26 , a uniform force @xmath4 from the external electric field , and a force @xmath27 from the impurity pinning .
this impurity force has an explicit dependence on the position @xmath23 , arising from the quenched randomness in the location of the impurities , through the functions @xmath28 and @xmath29 . these correspond physically to the strength of the impurity pinning , @xmath28 , and a preferred phase @xmath29 selected by the impurities .
( in order to fix the normalization of @xmath30 , we choose @xmath31 to have a maximum value of 1 . ) since the impurities have only short range correlations , @xmath28 and @xmath29 are taken to be uncorrelated from one position to another , with a distribution @xmath32 for @xmath30 and a uniform distribution over @xmath33 for @xmath34 .
the pinning force is the derivative , @xmath35 of an impurity pinning potential .
because of the @xmath17 translational invariance of the cdw , the force @xmath36 has to be @xmath17-periodic in @xmath37 .
among the quantities of physical interest in this model below threshold are the polarization , @xmath38 , defined as the mean displacement from some initial configuration of the phase @xmath37 ; close to threshold , the singular part of @xmath38 scales as @xmath39_s \sim(f_t - f)^{-\gamma+1}.\label{polarz}\ ] ] the derivative of this with respect to @xmath4 is the upwards polarizability , defined earlier as the response to an infinitesimal increase in @xmath4 , which scales as @xmath40/\delta f\sim ( f_t - f)^{-\gamma}.\label{suscep}\ ] ] in ref . , the behaviour above threshold for models of this type was analyzed .
this was done by expanding around the solution of eq.([flr ] ) within a mean field approximation , which is obtained by replacing the short - ranged elastic term @xmath41 with an infinite - ranged interaction , @xmath42 .
the mean field solution is known to depend on the details of the pinning potential @xmath43 : for instance , the velocity above threshold scales as @xmath44 in the critical regime,@xcite with an exponent @xmath45 for smooth pinning potentials@xcite and @xmath46 for pinning potentials which have wedge - shaped linear cusped maxima.@xcite this lack of universality arises from the fact that , for smooth pinning potentials , when the phase @xmath37 at any site passes through an unstable point in its local effective potential , it spends a long time accelerating before jumping forward to a new stable state .
this results in a second time scale that diverges at threshold ( in addition to the natural scale of @xmath47 ) .
no such second divergent time scale is present for linear cusped potentials , for which the acceleration time remains of @xmath48 arbitrarily close to threshold . despite this difference in behaviour for the two types of potentials in mean field theory ,
numerically it is found that the critical behaviour is _ independent _ of the shape of the pinning potential for @xmath49 , and is in fact even the same for automaton models . in ref . , it was argued that this is because the dynamics for small @xmath12 are very irregular : when the phase at any site jumps forward , it produces a sharp force on all the neighbours to which it is elastically coupled . for large @xmath12 ,
the total elastic force acting on any phase is the sum of contributions from a large number of neighbours ; however , as @xmath12 is decreased , the elastic force becomes more and more jerky .
for sufficiently small @xmath12 , the phase @xmath37 of the cdw at any site is ` kicked ' by its neighbours over the maximum of the local effective potential it experiences .
the details of the pinning potential are no longer important , and the second divergent time scale is eliminated , restoring universality .
the expansion around mean field theory was accordingly carried out for linear cusped potentials , for which the second time scale is absent initially , so that the expansion is better behaved .
when @xmath5 is approached monotonically from below threshold , the critical behaviour is controlled by large avalanches , in which the cdw phase in localized regions moves forward abruptly in response to a small change in the field . since the collapse to universality above threshold
was understood in terms of the jerky dynamics , the same should be true even below threshold .
this is indeed found to be the case numerically .
thus even below threshold , we seek to expand around the mean field solution for linear cusped potentials . before proceeding with the calculations ,
it is necessary to consider a possibility that might make this procedure erroneous .
while we have answered the question of history dependence by choosing a specific approach to threshold , namely increasing @xmath4 monotonically , this does not completely eliminate problems that arise from the existence of many metastable states .
an expansion around the mean field solution of the type we develop here corresponds to first increasing the force within an infinite ranged model , and then tuning the elastic coupling to a short ranged form ( _ i.e. _ , restoring the @xmath41 term in eq.([flr ] ) which was replaced by @xmath50 in mean field theory ) . on the other hand
, the physically relevant critical behaviour corresponds to _ first _ tuning the elastic coupling to a short ranged form , and _ then _ increasing the force _ within _ a short ranged model . when there are many metastable solutions , as is the case below threshold , the two methods could in principle yield different critical behaviour . since the cdw phase at any site moves only forward when the force is raised monotonically , such problems from metastability could arise if reducing the range of the elastic interactions from a mean field theory were to result in the phases at different sites trying to move _
backwards_. while this will of course be true for _ some _ of the phases , for which the coupling to their neighbours tends to retard their motion , we shall see later in this paper that the polarizability exponent @xmath8 is in fact larger for @xmath49 than in mean field theory .
thus the dominant effect of reducing the range of the elastic interactions is to make the phases move _ forward _ , making the polarization ( and hence the polarizability ) more divergent than in mean field theory .
we therefore expect our procedure of expanding around the mean field solution to yield the correct critical behaviour .
the expansion around mean field theory is carried out following the prescription of sompolinsky and zippelius;@xcite the details are given in ref .. the method yields an impurity averaged generating functional for the correlation and response functions of the phase @xmath37 , of the form @xmath51&[&d\hat\phi]\exp\{-\int d^d x dt\,\hat\phi(x , t)[\partial_t-\nabla^2+r]\phi(x , t)+\hat\phi(x , t ) [ f - f_{mf}(\langle\phi(t)\rangle)]\qquad\nonumber\\ & - & \int d^dx dt_1 dt_2\,\hat\phi(x , t_1 ) \hat\phi(x , t_2 ) c[\langle\phi(t_1)\rangle+\phi(x , t_1)-\langle\phi(t_2)\rangle -\phi(x , t_2)]\}.\label{genfunc}\end{aligned}\ ] ] here @xmath52 and @xmath53 are dummy field variables , in terms of which the long wavelength low frequency forms of the truncated correlation and response functions of @xmath37 are given by @xmath54 where the left hand side represents the generalized response to a perturbing force @xmath55 added to the right hand side of eq.([flr ] ) , and the right hand side is the expectation value using the measure of eq.([genfunc ] ) .
@xmath56 is like a coarse grained version of the phase @xmath57 .
it gives the deviations of @xmath57 from its spatially averaged value , @xmath58 , and must satisfy the consistency condition @xmath59 , which determines @xmath60 .
various other terms , that do not affect the critical behaviour , have been suppressed in eq.([genfunc ] ) .
eq.([genfunc ] ) is similar to the generating functional obtained in ref . above threshold , except for new terms that we discuss in the next paragraph .
the function @xmath61 is the same as the mean field phase phase correlations as a function of time _ above _ threshold : @xmath62[\phi(t_2)-vt_2]\rangle$ ] .
below threshold , the spatial average of the phase , @xmath63 , replaces @xmath64 . for any value of @xmath58 ,
the function @xmath65 is the force that would result in an average displacement @xmath58 in the mean field approximation to eq.([flr ] ) , while the function @xmath4 is the force that would produce the same average displacement in the full short - ranged model of eq.([flr ] ) . as mentioned in the previous paragraph , the difference between these two , which is the coefficient of the second term in eq.([genfunc ] ) , can be found by the consistency condition @xmath59 .
this again is similar to the case above threshold . despite the similarities ,
there are important differences between eq.([genfunc ] ) and the corresponding generating functional above threshold@xcite .
there is a non - zero ` mass term ' in eq.([genfunc ] ) , @xmath66 , which results in a finite response of @xmath37 to a spatially uniform dc force . above threshold , where this term vanishes exactly due to time translational invariance of the system
, @xmath67 has a @xmath68 singularity , corresponding to a finite response in the _ velocity _ of @xmath37 .
the bare value of @xmath69 is the inverse of the mean field polarizability @xmath6 ( along the monotonically increasing path ) ; @xmath69 vanishes at threshold , although the manner in which it approaches zero at @xmath5 depends on the distribution @xmath32 of the pinning strengths .
the generating functional @xmath70 also differs from the form above threshold by various other terms that also vanish at threshold like @xmath71 does .
for instance , the coefficient of the second term , @xmath72 has corrections of @xmath73 ^ 2)$ ] .
these , however , do not affect the critical behaviour and are not shown in eq.([genfunc ] ) . also , it is only when the mean field correlation and response functions are evaluated within the monotonically increasing approach to threshold that we obtain the functional forms in eq.([genfunc ] ) that resemble those above threshold . for other ( non - monotonic ) approaches
, this will not be the case . above threshold , where the steady state behaviour is unique
, no particular path needs to be specified .
the variable @xmath63 in the arguments of the functions @xmath74 and @xmath61 in eq.([genfunc ] ) implicitly controls the dependence on @xmath4 .
it is necessary to increase @xmath63 _ adiabatically _ in order for the form in eq.([genfunc ] ) to be valid ; otherwise , there would be various transients that would have to be included in the equation .
since the increase of the force is much slower than all the dynamics of the cdw , the equal time response and correlation functions really give the _
static _ behaviour .
although it would seem that eqs.([genfunc ] ) and ( [ mapping ] ) can be used to obtain responses to time dependent perturbations , there is an implicit constraint that the perturbing force must satisfy for our entire approach to be valid : @xmath75 for @xmath76 .
this constraint prevents us from obtaining the ac response , @xmath77 .
we now carry out a renormalization group analysis of eq.([genfunc ] ) along the lines of ref .. we rescale space and time as @xmath78 and @xmath79 , and integrate out modes of all frequencies in the momentum shell between the ( rescaled ) momenta @xmath80 and @xmath81 , where @xmath81 is the upper cutoff in momentum .
the rescaling of the fields @xmath52 and @xmath53 are fixed in mean field theory by the requirement that all quadratic terms in the exponential have to be invariant under this transformation , except the @xmath71 term ; the variation of this last term is interpreted as a change in @xmath4 under renormalization . fixing the other quadratic terms in the exponential yields @xmath82 and @xmath83 , with the dynamic exponent @xmath14 fixed at 2 by comparing the scaling of the @xmath84 and @xmath85 terms .
the consistency of the mean field scaling is verified by ascertaining that all higher order operators are irrelevant ; this is indeed the case for @xmath86 , which is therefore the upper critical dimension .
this upper critical dimension is the same as for @xmath87 . at @xmath88 , the entire function @xmath89
$ ] becomes marginal . for @xmath49 ,
the critical behaviour is changed from its mean field form . as discussed in ref .
, for @xmath49 the function @xmath61 splits up into a constant part @xmath90 that controls the static distortions in the phase , and a functional operator @xmath91 that controls the dynamics .
the two of these decouple from each other , and scale differently . for the dynamics above threshold ,
where the steady state solution is periodic , it is necessary to preserve the invariance under the transformation @xmath92 . for @xmath93 , the upper critical dimension , fluctuations in @xmath37 scale in the same way as @xmath94 ; this then requires @xmath52 to be invariant under rescaling , as discussed in ref .. the operator whose coefficient is the constant part of @xmath61 , @xmath90 , is then _ relevant _ , leading to anomalous scaling of the static correlations . below threshold
, it is precisely these _ static _ correlations that we are interested in . since the mean phase @xmath58 reaches a constant value for any fixed value of @xmath95 in steady state ( with @xmath58 increasing with @xmath4 ) , the periodicity of the phase variable need not be preserved under the rescaling of the renormalization group .
accordingly , we choose a scaling for the fields @xmath53 and @xmath52 _ different _ from that of ref . :
@xmath96 the scaling of the @xmath53 field is fixed by requiring the invariance of the @xmath90 term , which is unrenormalized by loop corrections .
( this is because all loop terms involve differences between the function @xmath61 at two different values of its argument , which are not changed by an addition of a constant @xmath90 .
although it is conceivable that singular corrections could affect this result , we assume here that this is not the case . ) as discussed in ref . , any renormalization of the @xmath97 term from the operator @xmath61 ( as well as from higher order irrelevant operators ) must be of the form @xmath98 , and thus the @xmath99 term also has no loop corrections , fixing the scaling of @xmath52 .
the scaling of the @xmath66 term yields a renormalized polarizability @xmath100 . finally , the @xmath101 term , which _ does _ receive loop corrections , is relevant , so that the scaling of @xmath4 can also be obtained by power counting : @xmath102 , thus yielding the correlation length exponent @xmath103 and , from the scaling of @xmath104 , @xmath105 as mentioned in eqs .
( [ intronu ] ) and ( [ introgamma ] ) .
although it is possible to obtain these exponents directly from a very simplified model of cdws@xcite , the exponent @xmath14 _ can not _ be obtained from a similar analysis and , as we shall now see , is nontrivial . a more detailed discussion of ref .
is given in section v. the exponent @xmath14 is not affected by the difference between the scaling here and that of ref . ,
although under the scaling used here , the operator @xmath61 is dangerously irrelevant , flowing towards a singular function of zero amplitude , with _ constant _ second derivative @xmath106 .
( with the scaling of ref . , @xmath61 flows towards a regular function of constant amplitude , with the same second derivative . )
the @xmath107 result for @xmath14 , obtained in ref .
, depends on the second derivative of @xmath61 , and is given by @xmath108 the breakdown of two sided scaling for cdws , with @xmath109 below threshold , and @xmath110 above threshold , is now seen as the result of the different static and dynamic behaviour of the system , rather than being due to any fundamental difference between the properties above and below threshold . even above threshold ,
the static correlation length still exists , and was indirectly obtained in ref . as the length that controls the finite size scaling behaviour .
the static correlation function above threshold varies as a simple power law with distance , and does _ not _ show any crossover at the static ( or dynamic ) correlation length , so that the static correlation length can not be obtained directly .
we now compare these results with those from numerical simulations .
previous simulations have often been carried out with the continuous variable dynamics of eq.([flr ] ) , with @xmath23 discretized to a lattice .
however , it is possible to obtain much more accurate results with automaton models , which may be viewed as a singular limit of eq.([flr ] ) , in which the pinning potential at any site has very narrow and steep wells , so that @xmath111 , with @xmath112 an integer variable .
a further approximation is made by discretizing time;@xcite the dynamics of @xmath113 is then @xmath114+h(x)>0\qquad\qquad\nonumber\\ & = & m(x;t-1)\qquad\qquad { \rm if } f+\nabla^2[2\pi m(x;t-1)+\beta(x)]+h(x)\leq 0.\label{discrtzn}\end{aligned}\ ] ] here the @xmath85 operator is the discrete laplacian .
this automaton model ( and variants of it ) are numerically much more efficient than direct simulations of eq.([flr ] ) .
since for small @xmath12 , the critical behaviour is independent of the form of the pinning potential used , we shall rely mostly on the automaton model simulations in this section , referring to continuous dynamics only for comparison .
it is more convenient to express the dynamics in terms of a local `` curvature '' variable @xmath115 + { \rm int}[\{f+\nabla^2\beta(x)+h(x)\}/2\pi]\label{curvature}\ ] ] where the sum over @xmath116 is restricted to sites neighbouring @xmath23 .
the first term in this equation is the discretized form of @xmath117 . by construction
, @xmath118 is an integer valued variable . from eq.([discrtzn ] ) , we see that the dynamics can be expressed in terms of @xmath118 as @xmath119 here @xmath120 is the number of neighbours that the site @xmath23 has .
we can view this as a sandpile model@xcite if we treat @xmath118 as the height of sand at the site @xmath23 ; when this exceeds a preset threshold ( zero in eq.([sandpile ] ) ) one grain of sand falls off from @xmath23 to each of its neighbouring sites . as @xmath4 is increased , @xmath121 $ ] increases by 1 whenever its argument crosses an integer . from eq.([curvature ] ) , we see that increasing @xmath4 towards threshold is equivalent to dropping grains of sand randomly on different sites of the system .
there is , however , a slight difference in the manner in which sand is added as compared to standard sandpile models : since the randomness from @xmath122 is _ quenched _ , sand grains are dropped in a random order on the different sites of the lattice one by one , with _ no _ site being revisited twice in a cycle . when every site has been visited exactly once , the whole cycle is repeated in the same order .
in contrast to this , in standard sandpile models the addition of each sand grain is an independent random process , and there are no correlations between the sites chosen . for a sufficiently large system , where the repeating cycle is very long , it is reasonable to expect the difference between cyclic and uncorrelated addition of sand grains to be inconsequential.@xcite
another difference between cdws and usual sandpiles is that open boundary conditions are normally used for sandpiles , whereas periodic boundary conditions are used for cdws . with periodic boundary conditions , a grain of sand that exits the system at one boundary reenters at the opposite boundary .
again , when the system size is much larger than the characteristic avalanche size , this difference should not affect the behaviour .
@xcite in this section , we shall use numerical results from cdw and sandpile models , _ both _ with periodic boundary conditions , to compare with our analytical results .
comparisons with earlier simulations@xcite on sandpiles with _
open _ boundary conditions will be discussed in subsequent sections .
figure 1 shows a finite size scaling plot for the polarization @xmath123 ( measured from the initial state @xmath124 at @xmath125 ) for @xmath21 as a function of the driving force @xmath4 and the linear size of the system , @xmath126 . the finite size scaling of the polarization should be of the form @xmath127 where @xmath128 is the reduced force , @xmath129 .
the collapse to a scaling form in the numerical data shown in figure 1 is not very good .
this is probably due to the fact that the polarization in any state involves the polarizability integrated from the initial to the present state .
corrections to scaling might thus be expected to persist in the polarization until the system is very close to threshold .
figure 2 shows a similar scaling plot for the polarizability , obtained by a numerical derivative of the polarization .
the data here scale better when @xmath130 is large , but show a large scatter very close to threshold due to numerical uncertainties .
figure 3 shows a scaling plot for the polarization in the sandpile version of the dynamics of eq.([discrtzn ] ) ; the scaling form works much better here . from the data in figure 3 , we obtain estimates for the critical exponents : @xmath131 and @xmath132 , in agreement with eqs.([obtgamma ] ) and ( [ obtnu ] ) .
simulations with the continuous dynamics of eq.([flr ] ) have larger numerical uncertainties . as detailed in ref .
, a log - log plot of the polarization as a function of @xmath128 appears to yield an exponent @xmath133 , although the slight upward curvature seen as @xmath134 allows for the possibility that the asymptotic value of @xmath8 is indeed 2 , as given by eq.([obtgamma ] ) .
the collapse of the data to a single scaling function is also not very satisfactory , but can be used to obtain the correlation length exponent @xmath16 as @xmath135 .
a more accurate value for @xmath16 is obtained from the width of the distribution of threshold fields @xmath5 for various systems of size @xmath126 with different realizations of randomness , since one expects finite size effects to become important when the characteristic size of an avalanche is of the order of the size of the system .
fitting @xmath136 to the form @xmath137 yields @xmath138.@xcite the results are thus consistent with eq.([obtnu ] ) . for three dimensional systems ,
the numerical uncertainties in the behaviour of the polarizability are larger . here
we show only the results for simulations on the sandpile model ( with periodic boundary conditions ) for which , as in two dimensions , the scaling is better .
figure 4 shows a scaling plot of the polarizability , from which we obtain @xmath139 and @xmath140 .
since it is possible to determine the location of the threshold field very accurately , one can also plot the polarization _ at _ threshold as a function of system size . as shown in figure 5
, @xmath141 appears to scale as @xmath142 , consistent with eqs .
( [ obtnu ] ) and ( [ obtgamma ] ) for @xmath143 .
( the error bars shown in figure 4 are from the statistical scatter in the data ; owing to the small range in @xmath126 , this is too conservative an estimate for the true uncertainties . )
the exponent @xmath16 can be obtained independently from the scatter in the threshold field to be @xmath144 , as shown in figure 6 . from the duration of an avalanche as a function of its size
, we obtain the dynamic exponent @xmath14 . the size @xmath145 of an avalanche , defined as the total number of sites participating in it , scales as @xmath146 , where @xmath147 is its linear extent , while @xmath14 is defined through @xmath148 , so that @xmath149 .
figures 7(a ) and 7(b ) are log - log plots of @xmath150 versus @xmath145 for cdws in 2 and 3 dimensions respectively . from these
we obtain @xmath151 and @xmath152 .
this agrees very well with our result @xmath153 ; @xmath154 can be found from the velocity exponent @xmath34 through @xmath155 , and @xmath34 is known independently from numerical simulations above threshold to be @xmath156 and @xmath157.@xcite
we now turn to the distribution of avalanches of different sizes in the critical region . as discussed earlier , there are differences in the forms for the continuous variable and automaton models .
this is because the total increase in the phase @xmath37 at any site is bounded above by @xmath17 in the continuous variable models .
no such constraint exists for automata , for which several avalanches of roughly the same size are grouped together into a single large avalanche .
this may lead to different number distributions for avalanches in the continuous and automaton models .
we first deal with the avalanche distribution for automaton models , returning later in this section to continuous variable models .
we conjecture a scaling form for the number distribution of avalanches@xcite : @xmath158 where @xmath159 is the total number of avalanches of size between @xmath145 and @xmath160 that occur in a unit volume of the system when the reduced force @xmath161 is changed from @xmath128 to @xmath162 .
this equation defines the exponent @xmath163 .
( here we have assumed that the same @xmath16 is associated with the characteristic size of the avalanches as well as the finite size effects . )
we define the moment of an avalanche as the total change in polarization that results from it .
this is related to the change in the phase at every site in the avalanche by @xmath164 . for continuous dynamics , where @xmath165 for all the sites , the moment scales as the size of the avalanche . for automaton models , the two are related by a nontrivial exponent :
@xmath166 equations ( [ avldistr ] ) and ( [ moment ] ) can be used to obtain an exponent identity relating @xmath163 and @xmath167 to @xmath8 . the total change in polarization upon changing the force by @xmath168
is found by integrating eq.([avldistr ] ) to be of the form @xmath169 . comparing with eq.([suscep ] ) , we obtain@xcite @xmath170 substituting eqs.([obtgamma ] ) and ( [ obtnu ] ) in this equation yields @xmath171 figure 8(a ) shows a finite size scaling plot of the distribution of avalanches for the automaton model in two dimensions . by fitting to the form of eq.([avldistr ] ) , we obtain @xmath172 figure 8(b ) shows a similar plot for @xmath143 , from which we can obtain @xmath173 the values of @xmath16 that are obtained from these scaling plots are @xmath132 and @xmath174 , consistent with eq.([obtnu ] ) ( as well as the scatter in @xmath5 ) .
the scaling of the moment of an avalanche with its size can be used to obtain numerically @xmath175 and @xmath176 which , together with the numerical values for @xmath163 , are consistent with eq.([gamma ] ) . in the limit as @xmath177 , a power law distribution for the avalanches
is obtained .
if @xmath178 is not a singular function of its argument @xmath23 for small @xmath23 , the limiting form of the distribution can be used to obtain @xmath163 and @xmath167 .
if , however , @xmath178 scales as @xmath179 for small @xmath23 , the apparent power law in the avalanche distribution would change from @xmath163 to @xmath180 , while leaving @xmath167 unchanged . a singular form for
@xmath181 would imply that the _ total _ number of avalanches generated by a small increase in @xmath4 ( which is dominated by small avalanches ) would be singular as @xmath5 is approached ; we have verified numerically that this is not the case for the automaton models .
it is interesting to note that eq.([gamma ] ) has been derived earlier by majumdar and dhar@xcite for sandpile models , by a completely different method that relies only on properties of clusters _ at _ the critical point.@xcite since sandpiles at the critical point are current driven , @xmath182 , instead of being tuned by a driving force @xmath4 , any prefactor in @xmath183 that is singular in @xmath128 would not be seen in the scaling .
the fact that eq.([gamma ] ) and the result of majumdar and dhar are identical can thus be taken as a proof that @xmath181 is not singular .
our results in eqs .
( [ k2 ] ) through ( [ th3 ] ) are also consistent with numerical simulations on sandpile models in two and three dimensions.@xcite as mentioned in the previous section , the dynamic exponent @xmath14 can be found numerically from the scaling of the duration of an avalanche with its size ; the results in both two and three dimensions are fairly close to a first order truncation of the @xmath2-expansion result of eq.([obtz ] ) . even at the critical point
, the duration of the avalanches can be used to obtain @xmath14 ; this allows us to compare with the result for two - dimensional sandpiles@xcite that @xmath184 , which agrees with our numerical results . for three dimensions ,
there is no exact calculation of @xmath14 at present to compare our numerical results with .
however , previous simulations on ( a transformed version of ) the sandpile model with open boundary conditions@xcite have yielded @xmath185 , in agreement with our results . for continuous variable models
, we have seen that the bound on the change in phase at any site in an avalanche constrains @xmath186 to being 1 .
( the subscript @xmath187 here denotes continuous variable dynamics . ) from eq.([gamma ] ) , we then obtain @xmath188 , so that both the exponents involved in the avalanche distribution are determined.@xcite this result is only true if the driving force is increased _ infinitesimally_. in practice , the force is increased in small steps ; no matter how small the step size , one presumably eventually crosses over into a regime where successive avalanches at the same site trigger within a single force step , and the exponents change to those of the automaton models .
although the analytical treatment of section iii is applicable only for continuous variable dynamics , a proper analysis of the singular part of the _ non - linear _ response to finite increases in @xmath4 might yield the automaton values for the exponent @xmath167 ; we have so far been unable to carry this out .
note that the numerical results in eqs.([k2 ] ) through ( [ th3 ] ) are not very far from the trivial values @xmath189 and @xmath190 ; in fact , in three dimensions they are consistent to within the numerical uncertainties . instead of triggering avalanches by increasing the driving force ,
it is possible to do so by applying small random ` kicks ' to the _ phases _ at different sites , which may be considered to mimic the effect of thermal noise .
remarkably , for small thermal noise , the characteristic linear extent of an avalanche still scales with the distance from the zero temperature threshold force , but with a _ different _ correlation length exponent : @xmath191 with @xmath192 .
the ` thermal ' correlation length exponent has been found numerically by myers@xcite to be @xmath193 in two dimensions . as shown in figure 9
, the same result is true for @xmath143 as well
. a proper understanding of this phenomenon is still lacking ; however , it is possible that considerations about the equilibrium behaviour ( at finite temperature ) in the _ absence _ of a driving force may be applicable here . at low temperatures
, one might expect the presence of a driving force to produce an overall forward drift of the cdw , while still preserving the equilibrium properties locally . in equilibrium
, the dynamics can be expressed in terms of correlations in the pinning potential .
this is in contrast to our zero temperature analysis , where the physically relevant correlation function @xmath61 is related to force - force correlations.@xcite the potential - potential correlation function is related to the function @xmath61 integrated twice ; as shown in ref .
, in equilibrium it is necessary for this to be periodic in its argument .
changing @xmath61 by a constant @xmath90 is therefore no longer allowed in equilibrium . setting @xmath90 to zero implies that the static correlations no longer scale differently from the dynamic correlations , and @xmath194 .
further work on this , especially on crossovers from equilibrium to nonequilibrium properties , is clearly required .
most of the work on sandpile models has been on the properties at the critical point . however , tang and bak@xcite have considered sandpile models away from the critical point ; through a combination of scaling arguments and numerical simulations , they have obtained results similar to some of ours .
for instance , by considering the diffusive nature by which sand grains propagate through the system , they obtain the exponent identity @xmath195 , which is the same as from eqs.([obtgamma ] ) and ( [ obtnu ] ) . with the assumption that the distribution of avalanches is independent of system size for sufficiently large systems , and the implicit assumption that the moment of an avalanche is proportional to its size , they obtain an exponent identity connecting @xmath8 and @xmath163 , which is a special case of the general result derived in the previous section .
the first of these two assumptions is equivalent to @xmath181 not being singular ; we have seen in the previous section that this is correct , both by comparing our scaling law to that of majumdar and dhar@xcite and by direct numerical verification .
the second assumption , that @xmath189 , is _ not _ correct , even though it is approximately valid for @xmath21 which is the case tang and bak focus on ( and actually much better for @xmath143 ) . a few other results that they obtain are for the behaviour of transients , and can not be compared with our work .
the authors have also noted that if the critical behaviour in these nonequilibrium models were to be like that of equilibrium statistical mechanical systems , various exponent identities relating @xmath34 to other quantities would result .
these exponent identities can be compared with our results , and are found _ not _ to be valid . surprisingly , the value obtained by them for @xmath16 through numerical simulations is @xmath196 , which disagrees completely with our analytical and numerical result @xmath197 .
it is not clear what this difference is due to , although a similar value of @xmath16 ( @xmath198 ) has apparently been observed in other simulations as well in which the sandpile is perturbed in a manner different from ours@xcite .
as discussed in the previous section , with different types of kicks on the sandpile ( that may or may not have physically meaningful analogs in the continuous variable models ) @xmath16 can be very different ; this might be the source of the discrepancy .
our simulations on sandpile models , detailed in the previous section , yield @xmath199 . a related problem to cdws , that of the depinning of interfaces in random media , has been analyzed in the moving phase by methods similar to those used for cdws.@xcite unlike the case for cdws
, the statics and dynamics do not behave differently ; physically , this is because as the interface moves forward it experiences different regions of impurities and is thus unable to build up anomalous static correlations .
formally , this is seen by the absence of an unstable direction at the fixed point of @xmath61 when the appropriate boundary conditions are taken . since this separate scaling of static and dynamic correlations is the source of the difference between @xmath200 and @xmath201 for cdws , two - sided scaling will be valid for interfaces .
the relation @xmath195 can be verified to be still true by the procedure in this paper .
parisi and pietronero@xcite have considered a model for cdws which is equivalent to choosing the pinning potentials @xmath43 at different sites to be of the form @xmath202 , so that the phase at any site experiences a _ constant _ retarding force from the pinning potential , equal to @xmath203 .
in addition to this , there is the elastic force from the neighbouring sites , and the external driving force .
( although they have a @xmath204-function constraint that prevents the total force at any site from being negative , we have seen that when threshold is approached monotonically the phases at all the sites always move forward , so that this constraint is inconsequential for the critical behaviour . )
although this model correctly yields @xmath205 and @xmath109 , the dynamic exponent @xmath14 is wrong , being trivially equal to 2 .
the reason for this partially correct result can now be understood : after averaging over the randomness , their model corresponds to the one we have considered here , with the dynamic part of the correlation function , @xmath91 , set to zero , but with @xmath206 .
this _ does _ constitute a fixed point , at which loop corrections from @xmath91 are absent , so that @xmath207 .
however , this fixed point is unstable to the one that we have analyzed in this paper , so that the model that they have considered constitutes a pathological limit for pinning potentials . in particular
, it is possible to replace the vertical drop in @xmath43 at @xmath208 with a very steep section of non - zero width .
the critical behaviour of this model has been found numerically@xcite to the same as for the other models that we have considered here , although the width of the critical region above threshold can be seen through a perturbation expansion to vanish as the width of the steep sections grows smaller.@xcite as mentioned in section iii , due to the restriction that the driving force must be monotonically increasing in time , it is not possible to obtain the ac response @xmath77 from our analysis .
it is not clear at present if some variation of our approach may prove adequate for this purpose .
it is important to remember that the universality that exists for various different models with regard to the monotonic approach to threshold is considerably limited for @xmath77 even in low dimensions . in this paper , we have analyzed the critical behaviour of charge density waves as the depinning threshold is approached monotonically from below .
our success in this was due to the fact that , for this special path to threshold , the existence of many metastable states was argued not to invalidate the perturbation expansion we carry out .
although for any particular problem one must consider afresh whether such a perturbation expansion is valid , it may be possible to use similar techniques for other systems .
we would like to thank daniel fisher , satya majumdar and chao tang for useful discussions .
is supported by a junior fellowship from the society of fellows at harvard university , and wishes to acknowledge the hospitality of the institute for theoretical physics in santa barbara where part of this work has been completed .
this is only correct if one neglects various effects like thermal fluctuations and dislocations ; in real materials , these either round out the sharp phase transition or drive it first order .
e.g. _ , s.n .
coppersmith , phys .
rev . lett .
* 65 * , 1044 ( 1990 ) , phys .
rev . b*44 * , 2887 ( 1991 ) ; s.n .
coppersmith and a.j .
millis , phys .
rev . b*44 * , 7799 ( 1991 ) . ]
the discretization of time in unit steps results in a constant force being indistinguishable from an ac force whose frequency is an integer .
this gives rise to mode - locking in the sliding state , as discussed in ref .. however , the discreteness of time does not affect the statics .
this is the case for generic initial conditions for @xmath118 .
for instance , if the initial configuration is chosen to be @xmath124 , the second term in eq.([curvature ] ) results in @xmath118 being random .
however , if we choose @xmath118 to be identically zero initially , the behaviour with cyclic addition of sand grains is atypical . for @xmath87 , where an infinite avalanche runs across the entire system , the different boundary conditions give rise to very different behaviour .
the phase transition to a moving state that is seen with periodic boundary conditions is destroyed with open boundary conditions , where the system expels an appropriate number of sand grains to bring it back to its critical point .
note that both the quenched randomness in the addition of sand grains as well as the different boundary conditions imposed result in different _ finite size corrections _ for cdws than for sandpiles , even below threshold .
the notation they use is slightly different from ours : they use exponents @xmath210/d$ ] and @xmath211 . using the result from eq.([obtnu ] ) that @xmath109 , their result agrees with eq.([gamma ] ) only if @xmath212 .
( although their result is expressed for the specific case of @xmath21 , it can be generalized straightforwardly . )
note that @xmath213 is _ not _ equal to 1 ; thus the number of avalanches at a fixed value of @xmath128 scales with sizes with the exponent @xmath214/d=-(\alpha_c-1)/2d\nu$ ] .
it is this exponent that is denoted as @xmath163 in ref ..
|
the critical behavior of charge - density waves ( cdws ) in the pinned phase is studied for applied fields increasing toward the threshold field , using recently developed renormalization group techniques and simulations of automaton models . despite the existence of many metastable states in the pinned state of the cdw
, the renormalization group treatment can be used successfully to find the divergences in the polarization and the correlation length , and , to first order in an @xmath0 expansion , the diverging time scale .
the automaton models studied are a charge - density wave model and a `` sandpile '' model with periodic boundary conditions ; these models are found to have the same critical behavior , associated with diverging avalanche sizes .
the numerical results for the polarization and the diverging length and time scales in dimensions @xmath1 are in agreement with the analytical treatment .
these results clarify the connections between the behaviour above and below threshold : the characteristic correlation lengths on both sides of the transition diverge with different exponents .
the scaling of the distribution of avalanches on the approach to threshold is found to be different for automaton and continuous - variable models .
|
the main purpose of this paper is to establish that there are no square @xmath0-flat sequences of polynomials of the type @xmath3 where for each @xmath4 and @xmath5 @xmath6 is the circle group .
it follows that there are only finitely many barker sequences .
we thus get an affirmative answer to turyn - golay s conjecture and erds s conjectures .
furthermore , our result implies that the spectrum of dynamical system arising from generalized morse sequences is singular for every continuous morse sequence .
this answer an old question due to m. keane @xcite .
0.1 cm we remind that turyn - golay s conjecture , arising from the digital communications engineering , state that the merit factor of any binary sequence is bounded .
+ the merit factor of a binary sequence @xmath7 is given by @xmath8 where @xmath9 .
clearly , turyn - golay s conjecture is equivalent to @xmath10 conjecture of erds which say that for any polynomial @xmath11 from the type we have @xmath12 for some absolutely constant @xmath13 .
this conjecture implies the well known ultraflat conjecture of erds which state that for any polynomial @xmath11 from the type we have @xmath14 @xmath13 . in the same spirit ,
newman mentioned ( without attribution ) that it is conjectured that there is a constant @xmath15 such that for any polynomial @xmath11 of type we have @xmath16 .
this conjecture is nowadays known as @xmath17 newman s conjecture .
it is obvious that newman s conjecture implies the two conjectures of erds s @xcite , ( * ? ? ?
* problem 22 ) , @xcite .
0.1 cm of course newman s conjecture implies also turyn - golay conjecture
. however , our arguments break down as far as the @xmath17-newman s conjecture is concerned .
+ we notice that our proof is based on the description of some arithmetic set associated to the sequence @xmath18 and on the exact computation of the @xmath19 norm of dirichlet kernel .
we further apply some tools from gowers s method @xcite .
+ p. erds wrote about his ultraflat conjecture in @xcite the following : + `` some of these questions may not be `` serious '' mathematics but i am sure the following final problem considered by d. j. newman and myself for a long time is both difficult and interesting : let @xmath20 is it true that there is an absolute constant @xmath21 so that for every choice of the @xmath22 @xmath23 this probably remains true if the condition @xmath24 is replaced by @xmath25 . ''
+ the last conjecture ( that is , when @xmath24 is replaced by @xmath25 ) was disproved by j - p .
kahane @xcite .
further , j. beck @xcite has shown that the sequence @xmath26 @xmath27 can be chosen to be flat in the sense of littlewood ( see definition below ) with coefficients chosen from the solutions of @xmath28 .
so beck s result naturally raises the question as to what is the smallest cyclic subgroup of the circle group which supplies coefficients @xmath29 s for a flat sequence in the sense of littlewood of trigonometric polynomials of the type . according to the extensive numerical computations described in @xcite ,
the flat polynomials in the sense of littlewood of the type may exist @xcite .
however , our arguments seems to be far form bring any contribution to this problem . although our result that there are no ultraflat polynomials of the type confirm the numerical evidence in @xcite .
+ let us further mention that several attempts have been made in the past to solve turyn - golay s conjecture or erds s conjectures . for a brief review on those attempts we refer the reader to @xcite and @xcite .
+ it turns out that turyn - golay s conjecture is related to some spectral problems in ergodic theory . indeed , t. downarowicz and y. lacroix established that turyn - golay s conjecture is true if and only if all binary morse flows have singular spectrum @xcite .
let @xmath30 denote the class of littlewood polynomials by which we mean the trigonometric polynomials of the type @xmath31 where @xmath32 , @xmath33 . note that we make first and last coefficient of @xmath11 positive in our definition .
this makes the correspondence @xmath34 defined below one - one .
let @xmath35 denote the class of newman - bourgain polynomials , i.e. , polynomials @xmath36 of type @xmath37 where @xmath38 , @xmath39 or @xmath40 , and where @xmath41 is the number of non - zero terms in @xmath36 which is also the number of @xmath42 with @xmath43 . note that if @xmath11 is as in and if we put @xmath44 then the polynomial @xmath45 is in class @xmath35 , where @xmath41 is the number of @xmath46 which is also the number of @xmath47 .
+ it is obvious that the finite sequence @xmath48 defined a subset of the set @xmath49 which we denote by @xmath50 . by abuse of notation
we will denote also by @xmath50 the subset associated to the sequence @xmath51 .
+ as is customary , we denote by @xmath52 , @xmath53 the @xmath54-root of unity given by @xmath55 and by @xmath56 the cardinal of the set @xmath57 . for @xmath58 , the discret fourier transform of the finite set @xmath57 mod @xmath59 and its balanced function
are given by @xmath60 where @xmath61 or @xmath62 for its balanced function .
notice that in the usual sense @xmath63 is discret fourier transform of the function @xmath64 given by @xmath65 + the class of polynomials with coefficients of modulus one is denoted by @xmath66 . to avoid heavy notation , we denoted also by @xmath66 the class of @xmath0-normalized polynomials from class @xmath66 .
0.1 cm [ [ a - formula - between - littlewood - and - newman - bourgain - polynomials . ] ] * a formula between littlewood and newman - bourgain polynomials .
* + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + 0.1 cm let us define one - one invertible map @xmath67 from the class @xmath68 to the class @xmath35 by @xmath69 where @xmath70 , and @xmath41 is the number of @xmath43 which is also the number of @xmath47 .
+ note @xmath71 let @xmath72 we thus have that @xmath73 , while for @xmath74 , @xmath75 as @xmath76 .
+ the formula for polynomials in @xmath30 mentioned in the title of this subsection is as follows : if @xmath11 is as in then @xmath77 where @xmath41 is the number of terms in @xmath11 with coefficient + 1 , @xmath78 , and @xmath79 the proof follows as soon as we write @xmath80 and @xmath81 in the right hand side in full form and collect the coefficient of @xmath82 .
+ note that polynomials in @xmath30 , @xmath35 and the polynomial @xmath83 all have @xmath84 norm 1 .
moreover , for each @xmath85 , we have @xmath86 + by the identity .
0.2 cm we will further need the following fundamental inequalities from the interpolation theory due to marcinkiewz & zygmund ( * ? ? ?
* theorem 7.10 , chapter x , p.30 ) .
[ mz ] for @xmath87 , @xmath88 , and any analytic trigonometric polynomial @xmath11 of degree @xmath89 , @xmath90 where @xmath91 and @xmath92 are independent of @xmath93 and @xmath11 . for the trigonometric polynomials ,
marcinkiewz - zygmund interpolation inequalities can be stated as follows ( * ? ? ?
* theorem 7.5 , chapter x , p.28 ) .
[ mz2 ] for @xmath87 , @xmath88 , and any trigonometric polynomial @xmath11 of degree @xmath89 , @xmath94 where @xmath91 and @xmath92 are independent of @xmath93 and @xmath11 .
we will also need the following special case of marcinkiewicz - zygmund inequalities for the @xmath0 and @xmath10 norms .
we include the proof for sake of completeness .
[ classical ] let @xmath11 be an analytic trigonometric polynomial with degree @xmath95 and complex coefficients @xmath96
. then @xmath97 assume further that the degree of @xmath11 is odd and the coefficients are real , then @xmath98 an easy computation gives , for any @xmath99 , @xmath100 we thus get @xmath101 since @xmath102 the identity follows once we observe that @xmath103 for the proof of , we rewrite as follows @xmath104 where @xmath105 are the autocorrelations of the sequence @xmath106 .
since the coefficients are real , take the following form .
@xmath107 hence @xmath108 proceeding in the same manner as before we obtain @xmath109 and @xmath110 since @xmath54 is odd , by adding to we get .
this finishes the proof of the lemma .
\(1 ) the key argument in the proof of lemma [ classical ] can be reformulated as follows : let @xmath111 and @xmath112 then , for all @xmath113 , we have @xmath114 that is , the family @xmath115 is an orthonormal basis for @xmath116 .
we further notice that @xmath117 is an orthonormal basis for @xmath118 .
moreover , it is well known that the previous arguments are at the heart of the fast fourier transform ( fft ) techniques , we refer to ( * ? ? ? *
chap.7 ) for more details . for the application of fft in a variety of areas including biomedical engineering and the radar communications field , please refer to ( * ? ?
* chap . 8) and the references given there
. 0.1 cm ( 2 ) let us further point out that the identities and can be obtained as an easy application of parseval identity .
indeed , write @xmath119 and apply the parseval identity to obtain @xmath120 an elementary calculation show that this last identity is exactly the identity . for the second identity ,
the detailed verification is left to the reader .
[ [ flat - polynomials . ] ] * flat polynomials . * + + + + + + + + + + + + + + + + + + + for any @xmath121 or @xmath122 , the sequence @xmath123 of analytic trigonometric polynomials of @xmath124 norm 1 is said to be @xmath125-flat if @xmath126 converges in @xmath125-norm to the constant function @xmath127 . for @xmath128 , we say that @xmath129 is @xmath125-flat , if the mahler measures @xmath130 converge to 1 .
we recall that the mahler measure of a function @xmath131 is defined by @xmath132 the sequence @xmath123 is said to be square @xmath125-flat if @xmath133 converges in @xmath125-norm to the constant function @xmath127 .
+ obviously , if the sequence @xmath123 is square @xmath125-flat then it is @xmath125-flat , since @xmath134 and @xmath135 we further have that the square @xmath17-flatness is equivalent to the @xmath17-flatness , by proposition 4.2 from @xcite .
+ we say that the sequence @xmath123 is flat in almost everywhere sense ( a.e .
@xmath136 ) if @xmath137 , @xmath138 converges almost everywhere to @xmath127 with respect to @xmath139 . + following @xcite
, the sequence @xmath140 of polynomials from the class @xmath141 ( or @xmath66 ) is flat in the sense of littlewood if there exist constants @xmath142 such that for all @xmath143 , for all @xmath144 ( or at least for a large @xmath93 ) @xmath145 it is immediate that the flatness properties are invariant under @xmath146 .
it is further a nice exercise that the @xmath10 conjecture of erds and the ultraflat conjecture of erds holds in the class of newman - bourgain polynomials @xcite , @xcite . + we will need the following fundamental criterion of the square @xmath0-flatness .
[ cri1]let @xmath147 be a sequence of @xmath0-normalized polynomials .
then , @xmath147 is square @xmath0-flat if and only if the @xmath10-norm of @xmath148 converge to @xmath127 as @xmath149 .
a straightforward computation gives @xmath150 therefore , @xmath151{}}0,\ ] ] if and only if @xmath152{}}1.\ ] ] the proof of the proposition is complete .
0.5 cm we are now able to state our first main result .
[ mainl1 ] there no square @xmath0-flat polynomials sequence from the class of littlewood polynomials . from this
, it follows that the @xmath10-conjecture and the ultraflat conjecture of erds holds for the class of littlewood polynomials .
before stating our second and third main results , we need to recall the notion of barker sequences and some basic facts on the notion of dynamical systems arising from generalized morse sequences . + [ [ baker - sequences - and - the - connection - to - digital - communications - engineering . ] ] * baker sequences and the connection to digital communications engineering . * +
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + barker sequences are well - known in the streams of investigation from digital communications engineering .
barker introduced such sequences in @xcite to produce a low autocorrelation binary sequences , or equivalently a binary sequence with the highest possible value of @xmath153 .
the largest well - known values of @xmath153 are @xmath154 and @xmath155 obtain respectively by the following sequences @xmath156 and @xmath157 no other merit factor exceeding @xmath158 is known for any @xmath93 .
it was conjectured that @xmath159 and @xmath160 are the maximum possible values for @xmath153 .
this conjecture still open .
+ given a binary sequence @xmath161 , that is , for each @xmath162 , @xmath163 .
the @xmath164-th aperiodic autocorrelation of @xmath165 is given by @xmath166 for @xmath167 we put @xmath168 .
@xmath165 is said to be a barker sequence if for each @xmath169 we have @xmath170 the barker sequences and their generalizations have been a subject of many investigations since 1953 , both from digital communications engineering view point and complex analysis viewpoint .
therefore , there is an abundant literature on the subject , we refer to @xcite , @xcite , @xcite , @xcite , @xcite and the references therein for more details . here
, we remind only the following result need it .
[ barker ] let @xmath171 be a barker sequence with length @xmath93 . 1 .
if @xmath93 is odd then @xmath172 , if not and @xmath173 then @xmath174 for some integer @xmath41 .
2 . assume further that there exist a barker sequence with arbitrary length and let @xmath175 be a littlewood polynomial whose coefficients form a barker sequence of length @xmath93 .
then the sequence @xmath129 is square @xmath0-flat .
\(1 ) is due to turyn and storer @xcite .
the second part ( 2 ) is essentially due to saffari @xcite , we refer also to the proof of theorem 4.1 in @xcite line 5 . 0.1 cm at this point ,
let us state our second main result .
[ mainl2 ] there are only finitely many barker sequences .
[ [ weak - banach - rohklin - spectral - problem - and - the - connection - with - ergodic - theory . ] ] * weak banach - rohklin spectral problem and the connection with ergodic theory . *
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in @xcite bourgain showed that the @xmath17-flat polynomials problem in the class @xmath176 is related to the nature of the spectrum of the class 1 maps introduced by ornstein @xcite .
this class of maps is nowadays called rank one maps and there is a large literature on it .
later , m. guenais in @xcite established that @xmath17-flat polynomials problem in the class @xmath141 is equivalent to the weak banach - rohklin problem in the class of morse coycle extension maps .
+ following ulam the banach problem from the scottish book can be stated as follows@xcite .
does there exist a square integrable function @xmath177 and a measure preserving transformation @xmath178 , @xmath179 , such that the sequence of functions @xmath180 forms a complete orthogonal set in hilbert space ? 0.1 cm obviously , banach problem has a positive answer in the class of non - conservative dynamical systems .
for the conservative case , the problem remained open until very recently , when it was answered affirmatively by the author in @xcite .
precisely , therein , the author produced a conservative ergodic infinite measure preserving with simple lebesgue spectrum . for more details , we refer to @xcite .
+ the russian related problem to the banach problem is known as rohklin problem .
rohklin asked in @xcite on finding a map acting on the finite measure space with finite lebesgue spectrum .
to the best knowledge of the author , this problem still open . however , the weak rohklin problem on the existence of dynamical system with finite lebesgue component was solved since 1982 by j. mathew and m. g. nadkarni @xcite , t. kamae @xcite , m. queffelec @xcite , and o. ageev @xcite .
indeed , the authors produced a dynamical system with lebesgue component of multiplicity two .
fifteen years later , m. guenais produced a torsion group action with lebesgue component of multiplicity one @xcite .
0.1 cm historically , the problem on finding a map acting on a probability space with simple lebesgue spectrum seems to be initiated by banach and rohklin . on this problem , and
more generally , on rohklin problem , kirillov in his survey paper @xcite wrote there are grounds for thinking that such examples do not exist " .
the weak banach - rohklin problem raised the question of whether there exists an ergodic map acting on a probability space with simple lebesgue component plus some singular part in the spectrum . if we require only that the map is non - singular and ergodic then , very recently , m. nadkarni and the author established that the problem has an affirmative answer @xcite .
0.1 cm here , we will summarize briefly the connection between the square @xmath0-flatness and the so - called morse dynamical systems .
0.1 cm the one - sided generalized morse sequence @xmath57 is defined as the coordinatewise limit of the blocks @xmath197 notice that convergence is granted by the condition . + the one - sided generalized morse sequence @xmath57 is extended in the usual manner to be the bi - infinite sequence which we still denoted by @xmath57 .
+ 0.1 cm let @xmath146 be a shift map on the compact space @xmath198 given by @xmath199=x[n+1]$ ] , @xmath200 , and @xmath201 , be a generalized morse sequence .
the morse flow is defined as the subshfit generated by @xmath57 , that is , the topological dynamical system @xmath202 where @xmath203 is the closure of the orbit of @xmath57 under @xmath146 .
+ in the language of cocycles , the morse flow can be defined as @xmath204-point extension of the odometer .
the associated cocycle @xmath205 is defined inductively , and it is continuous at all points except at one point .
@xmath205 is also called a continuous morse cocycle .
+ the generalized morse sequences and morse cocycle has been the subject of many investigations and publication , for more details we refer to @xcite , @xcite and the references therein . + we remind that m. guenais in @xcite established that the morse cocycle has a simple lebesgue spectrum if and only if there exists a sequence of @xmath17-flat polynomials . + 0.1 cm for a continuous cocycle , downarowicz - lacroix in @xcite proved the following : [ dow - l ] the dynamical system arising from the continuous morse sequence @xmath206 has a simple lebesgue component if and only if the polynomials @xmath207 are square @xmath0-flat . 0.2 cm our third main result concern the spectrum of dynamical system arising from the generalized morse sequences , and it can be stated as follows [ mainl3 ] the spectrum of any morse flow arising from continuous morse sequence is singular .
we start by recalling the following special case of the @xmath125-flatness criterion from @xcite .
we include the proof for sake of completeness .
let @xmath208 and @xmath147 be a sequence of @xmath0-normalized polynomials such that for each @xmath54 the degree of @xmath148 is @xmath209
. then @xmath210 are square @xmath125-flat if and only if @xmath211{}}0,\ ] ] and @xmath212{}}0.\ ] ] [ appmz ] let @xmath147 be a sequence of @xmath0-normalized polynomials such that for each @xmath54 the degree of @xmath148 is @xmath209 .
then @xmath214 are square @xmath0-flat if and only if @xmath215{}}0,\ ] ] and @xmath216{}}0.\ ] ] assume that there exists a sequence @xmath220 of @xmath0-normalized littlewood polynomials such that the @xmath10-norm @xmath220 converge to 1 as @xmath221 .
furthermore , without loss of generality , assume that the degree of @xmath220 are odd .
then , by appealing to proposition [ cri1 ] , the sequence of the polynomials @xmath220 is square @xmath0-flat .
we thus have @xmath215{}}0,\ ] ] by corollary [ appmz ] . combining this with , we obtain @xmath222{}}0.\ ] ]
consequently , @xmath223{}}0.\end{aligned}\ ] ] now , applying lemma [ classical ] , we get @xmath224{}}0 , \nonumber\end{aligned}\ ] ] we further have @xmath225 and @xmath226 moreover , for any @xmath99 , @xmath227 where @xmath105 are the autocorrelation coefficients of @xmath228 these autocorrelation coefficients are also called _ the autocorrelation coefficients of @xmath50 . + integrating , we get @xmath229 we can thus rewrite as follows @xmath230 taking into account our assumption , we see by proposition [ jjh ] that @xmath231{}}-1.\ ] ] this allows us to assert that the sequence @xmath232 + converge to @xmath127 as @xmath149 .
+ at this point , we claim that for @xmath54 large enough , @xmath233 to see this , write @xmath234 and by appealing to the triangle inequality , we get @xmath235 moreover , a straightforward computation gives @xmath236 and @xmath237 therefore , for @xmath54 large enough , @xmath238 combining this with our assumption , we obtain @xmath239{}}1\ ] ] and @xmath240{}}1.\ ] ] whence , by , @xmath241{}}1,\ ] ] which ends the proof of the claim .
+ applying lemma [ classical ] again , it follows that @xmath242 hence @xmath243 therefore @xmath244 we further notice that for @xmath54 large enough @xmath245 at this point we claim that for @xmath54 large enough we have @xmath246 indeed , by the well - known lagrangian interpolation formula , we have @xmath247 whence @xmath248 the second identity follows from .
we thus get from the following identity @xmath249 applying the triangle inequalities we obtain @xmath250 and @xmath251 moreover , by taking into account the following estimation from @xcite , ( * ? ? ?
* art.67,68 ) , @xmath252 where @xmath253 .
the correlation coefficients are the autocorrelation of the binary sequence @xmath254 .
it can be also obtained by applying to the dirichlet kernel . ]
@xmath255 we see that @xmath256 that is , @xmath257 applying again lemma [ classical ] combinded with our assumption we infer that the sequence @xmath258 is bounded .
we thus have the following estimation @xmath259 which gives @xmath260 and this finish the proof of the claim .
+ we deduce that the autocorrelation coefficients @xmath105 of @xmath50 satisfy @xmath261 in other words @xmath262 since @xmath263 where @xmath264 is the standard convolution operation .
+ we further observe that the quantity @xmath265 is the number of the solution of the equation @xmath266 , @xmath267 . + in the same manner
we can see that the autocorrelation coefficients @xmath268 of @xmath269 satisfy @xmath270 where @xmath269 is the complement of @xmath50 .
it follows that for large enough @xmath54 @xmath271 where @xmath272 hence , by a variant of balog - szemerdi s theorem , we can describe approximately the structure of @xmath50 @xcite . as a consequence ,
we obtain a contradiction .
but , we can also proceed directly . indeed , for each @xmath273 , @xmath274 where @xmath275 is given by @xmath276 _
notice here that , without loss of generality , we can suppose @xmath54 is an even integer . otherwise , by appealing to we can write @xmath282 hence @xmath283 applying the same arguments as before , we see that holds with a suitable function @xmath275 .
+ we need now to estimate the @xmath10-norm of @xmath284 .
write @xmath285 and put @xmath286 we thus need to estimate only the @xmath10-norm of @xmath287 .
+ observe that the coefficients of the analytic polynomial @xmath288 are in @xmath289 .
furthermore , without loss of generality , we can assume that the frequence of @xmath290 , @xmath127 and @xmath291 exists .
for @xmath292 , we denotes by @xmath293 the frequence of @xmath294 .
+ therefore , as in the proof of proposition [ jjh ] , if @xmath295 and @xmath296 . then @xmath297{}}+\infty .
\eqno{(gjjh)}\ ] ] indeed , employing marcinkiewicz - zygmund inequalities , we see that @xmath298 moreover @xmath299 where @xmath300 , @xmath301 and @xmath302 is the complement of the set @xmath303 .
we thus get for @xmath54 large enough , @xmath304 consequently , @xmath305 letting @xmath278 , we obviously obtain @xmath306 .
we thus need to examine only the case @xmath307 . for that , in the same spirit as before , we complete the proof as follows .
+ employing marcinkiewicz - zygmund inequalities we can assert that @xmath308 therefore if the frequencies of odd integers and even integers in @xmath50 are not balanced then @xmath309 { } } + \infty.\ ] ] we thus get a contradiction . otherwise the frequencies are balanced and this also yields a contradiction .
indeed , let @xmath58 be an integer , and observe once again that we can assume that @xmath59 divides @xmath54 .
hence again by marcinkiewicz - zygmund inequalities , we see that it follows that if for some @xmath59 and @xmath311 , we have @xmath312}\big)(\ell)\big|>0.\ ] ] then @xmath309{}}+\infty,\ ] ] which gives a contradiction .
otherwise @xmath50 mod @xmath59 is almost equipped with the uniform probability measure of @xmath313 , which also yields a contradiction .
+ summarizing , we conclude that @xmath220 can not be square @xmath0-flat and the proof of the first main result is complete .
the proof of theorem [ mainl2 ] and [ mainl3 ] is straightforward from theorem [ mainl1 ] , since the spectrum of morse cocycle satisfy the purity law which say that the spectrum is either equivalent to the lebesgue measure on @xmath6 or singular to it . for the proof of this last fact
we refer to @xcite or @xcite .
\(i ) the estimation of @xmath10-norm of dirichlet kernel @xmath314 can be obtained as a corollary of the following more general estimation of @xmath19 norm of dirichlet kernel established in ( * ? ? ?
* lemma 3 ) @xmath315 indeed , for @xmath316 , we have @xmath317 \(ii ) notice that the argument @xmath318 generalized jensen - jensen and hholdt s result ( proposition [ jjh ] . ) .
we further obtain in our last argument a generalization of this result to the high degree . + ( iii ) some of our arguments are valid if we assume that @xmath220 is @xmath125-flat , @xmath208 . however , we need some new ideas to tackle the problem of @xmath125-flatness in the class of littlewood , for any @xmath208 . as a consequence ,
it is natural to ask the following question about the constant @xmath319 in the marcinkiewicz - zygmund inequalities the author wishes to express his thanks to franois parreau , jean - paul thouvenot , mahendra nadkarni , for many stimulating conversations on the subject and their sustained interest and encouragement in this work .
the author is specialy indebted to mahendra nadkarni for pointing to him a gap in the previous draft and for bring to his attention some deep remarks .
he further wishes to express his thanks to william veech and herv queffelec for their encouragement and interest in this work .
he would like also to thanks the erwin schrdinger institute and the organizers of the workshop
the workshop on normal numbers : arithmetic , computational and probabilistic aspects for the invitation .
e. h. el abdalaoui and m. nadkarni , _ a non - singular transformation whose spectrum has lebesgue component of multiplicity one , _ ergodic theory and dynamical systems , available on cjo2014 .
doi:10.1017/etds.2014.85 . http://fr.arxiv.org/pdf/1402.1312 .
p. borwein , r. ferguson and j. knauer , _ the merit factor problem _ , the merit factor problem .
number theory and polynomials , 52 - 70 , london math .
lecture note ser .
, 352 , cambridge univ . press , cambridge , 2008 .
p. borwein and m. j. mossinghoff , barker sequences and flat polynomials , number theory and polynomials ( bristol , u.k . , 2006 )
( j. mckee and c. smyth , eds . ) , london math .
lecture note ser .
352 , cambridge univ . press , 2008 , pp . 7188 .
p. erds , problems and results on polynomials and interpolation .
aspects of contemporary complex analysis ( proc .
study inst .
durham , durham , 1979 ) , pp .
383391 , academic press , london - new york , 1980 .
h. helson and w. parry , cocycles and spectra , ark . mat . , 16 ( 1978 ) , 195 - 206
. t. hholdt , _ the merit factor problem for binary sequences , _ applied algebra , algebraic algorithms and error - correcting codes , 5159 , lecture notes in comput .
, 3857 , springer , berlin , 2006 .
d. ornstein , _ on the root problem in ergodic theory _ proc .
sixth berkeley symposium on mathematical statistics and probability , berkeley and los angeles , uiversity of california press , * vol 2*(1970 ) , 348 - 356 .
k. r. rao , d. n. kim and j. j. hwang , fast fourier transform : algorithms and applications , signals and communication technology , springer , dordrecht , 2010 .
v. a. rokhlin , selected topics in the metric theory of dynamical systems , uspekhi mat .
nauk new series " , 4 ( 1949 ) , 57 - 128 ( russian ) ; amer .
transl . 2 , 40 ( 1966 ) , 171 - 240 . b. saffari & b. smith , _ inexistence de polynmes ultra - plats de kahane coefficients @xmath326
. preuve de la conjecture derds .
( french ) [ nonexistence of ultra - flat kahane polynomials with coefficients @xmath327 .
proof of the erds conjecture ] , _ c. r. acad .
paris sr .
306 ( 1988 ) , no .
16 , 695698 .
m. fredman , b. saffari & b. smith , _ polyn^ omes
rciproques : conjecture derds en norme @xmath10 , taille des autocorrlations et inexistence des codes de barker .
( french ) [ self - inversive polynomials : @xmath10-norm erds conjecture , size of autocorrelations and nonexistence of barker codes ] _ , c. r. acad .
paris sr .
i math . 308
( 1989 ) , no .
15 , 461464 .
barker sequences and littlewood two - sided conjectures on polynomials with @xmath327 coefficients , sminaire danalyse harmonique , anne 1989/90 , 139151 , univ .
paris xi , orsay , 1990 .
r. turyn and j. storer , on binary sequences , proc .
( 1961 ) , 394399 .
|
there are no square @xmath0-flat sequences of polynomials of the type @xmath1 where for each @xmath2 .
it follows that erds s conjectures on littlewood polynomials hold . consequently , turyn - golay s conjecture is true , that is , there are only finitely many barker sequences .
we further get that the spectrum of dynamical systems arising from continuous morse sequences is singular .
this settles an old question due to m. keane .
|
ru@xmath9o@xmath10 for fields @xmath5 applied with an angle @xmath11 versus the @xmath12-plane .
the surface represents first - order transitions separating a region with low ( spin ) polarization from a region with high polarization .
the thick black line connecting @xmath13 and @xmath14 is a line of critical endpoints .
for details see main text . ]
the ruthenium compounds of the ruddlesden - popper series , sr@xmath15ru@xmath16o@xmath17 , have been the subject of intensive research for over a decade due to their interesting ground - states .
while the quasi - two - dimensional single - layer ( n=1 ) is an unconventional superconductor likely with @xmath18-wave pairing,@xcite there is a growing tendency towards ferromagnetism with increasing layer number .
the three - dimensional infinite - layer compound , srruo@xmath0 , indeed realizes an itinerant ferromagnet .
@xcite the bilayer compound ( n=2 ) with its intermediate dimensionality , however , shows no ordering down to lowest temperatures@xcite and is supposed to be a strongly correlated fermi liquid .
still , applying uniaxial pressure can induce a ferromagnetic transition and this compound is , thus , expected to be on the verge to a ferromagnetic instability.@xcite this is also supported by inelastic neutron scattering@xcite and band structure calculations .
@xcite + as was first discussed by wohlfarth and rhodes,@xcite such proximity to ferromagnetism can result in metamagnetic behavior , a superlinear rise in the magnetization over a narrow region of applied magnetic field @xmath5 .
this phenomenon was observed in a number of systems @xcite and also in sr@xmath0ru@xmath1o@xmath2 it was found@xcite with a critical field @xmath3 , however , that depends strongly on the angle of the field versus the @xmath12-plane .
while for @xmath19 the critical field @xmath20 t , it rises to @xmath21 t for @xmath22 .
in addition , a first - order transition occurs for in - plane fields below @xmath23k , while there is only a crossover for fields parallel to @xmath6 .
it was therefore suggested that the field angle could be used as a tuning parameter for a line of first - order transitions that goes to @xmath24 around @xmath25 , thus , realizing a quantum critical endpoint @xcite ( see schematic phase diagram in fig .
[ fig : schem - phase ] ) .
+ however , when trying to reach this quantum critical endpoint on ultra - pure single crystals with residual resistivities down to @xmath26 cm , a splitting of the metamagnetic transition into two jumps was observed .
these jumps define an intermediate phase whose exact boundaries could be determined by measuring of several thermodynamic properties .
@xcite later , it was shown that this phase breaks the symmetry of the crystal@xcite and it was argued that this was due to an induced anisotropic electronic state with a symmetry - breaking fermi surface deformation similar to a pomeranchuck instability .
@xcite that this kind of a phase , also called nematic phase in analogy to liquid crystal phases , can lead to two consecutive metamagnetic transitions had already been shown in a paper by kee and kim .
@xcite in addition , the anomalous @xmath27 dependence of the susceptibility @xmath28 and the specific heat coefficient @xmath29 could be explained .
@xcite moreover , the two - fold degeneracy of the nematic phase allows for domain formation , such that domain - wall scattering could account for the increased resistivity of the intermediate phase .
@xcite + the electronic structure of a single layer of sr@xmath0ru@xmath1o@xmath2 is dominated by bands originating from the 4d @xmath30 orbitals @xmath31 , @xmath32 and @xmath33 hybridizing with the o 2p orbitals .
this leads in a simple approximation to two quasi - one - dimensional bands with mainly @xmath31 and @xmath32 character and a two - dimensional band stemming from the @xmath33 orbital .
these three bands are then additionally split due to the interlayer coupling resulting in 6 bands . an important consequence of the bilayer splitting is that one of the two bands coming from the d@xmath34 orbitals is shifted closer to the van hove singularity .
this was also confirmed by recent angle - resolved photoemission spectroscopy ( arpes ) measurements .
@xcite a chemical potential in the vicinity of a van hove singularity is the condition for the scenario described by binz and sigrist@xcite for a metamagnetic transition .
proximity to a van hove singularity can also lead to a nematic phase accompanying a metamagnetic transition as described by grigera et al .
@xcite the anisotropy in the critical field strength could then be explained by spin - orbit coupling ( soc ) effects similar to @xcite leading to an anisotropic effective @xmath35-factor .
+ a different route is taken by raghu et al .
@xcite and lee et al .
@xcite who studied a model where the metamagnetic transition comes from the two ( bilayer - split ) one - dimensional bands .
the anisotropy is then again introduced by considering spin - orbit interaction on the ru - sites and the nematic phase can be understood as an orbital ordering among these one - dimensional bands .
+ both these routes suffer , however , from a short - coming : even though they can describe the existence of a nematic phase and a dependence of @xmath3 on the field angle , they can not explain why the nematic phase occurs only for fields almost parallel to the crystalline @xmath6-axis .
+ we will address this point in the present work .
we study a model based on a two - dimensional band in a single layer originating from the @xmath33 orbitals .
the bilayer effects are only taken into account by placing this band closer to the van hove singularity .
starting from this , we will consider the effect of a lattice distortion in the planes .
the o - octahedra in sr@xmath0ru@xmath1o@xmath2 are rotated by @xmath36@xcite and we will show how this introduces a staggered spin - orbit coupling , an effect similar to the dzyaloshinski - morya interaction for localized spins , here , however , for itinerant electrons . for magnetic fields applied in the plane , this will add a component with wave vector @xmath37 to the static susceptibility . the induced commensurate spin - density wave ( sdw ) will open gaps in the fermi surface close to the van hove points which will have an impact on the occurrence of any instability that emerges due to the proximity to a van hove singularity . we choose here
the electronic nematic phase to examine this aspect , since it relies on the presence of the van hove singularity and represents one of the most promising candidates for the intermediate phase . in order to discuss the essential influence of the spin - orbit coupling on the phase diagram
we adopt here a mean - field approach , with the short - coming that critical fluctuations are not included well .
while , in particular , quantum critical fluctuations represent an intriguing part of the phenomenology of this metamagnetic transition , we assume that they are not essential to understand the basic effects due to spin - orbit coupling and lie beyond the scope of this study .
+ this paper is organized as follows : in section ii , we will introduce our model based on a three - band hamiltonian consisting of the ru 4@xmath33 orbital and the in - plane o 2@xmath18 orbitals . after reducing this model to an effective one - band model
, we will analyze the effect of a rotation of the oxygen octahedra on it . on - site interactions
are then treated within mean - field theory and an additional applied magnetic field is considered .
the resulting model is then studied in a next section .
following metzner et al.,@xcite we will in section iv add a forward - scattering term to allow for a nematic phase and study the influence of the staggered spin - orbit coupling to this phase . in a last section
we will discuss and summarize our findings . and o 2 p@xmath38 / p@xmath39 orbitals .
the relative signs of the hopping integrals come from the phases of the orbital wave - functions . ]
the starting point for our model is a three - band tight - binding model including the in - plane 4d ru orbital ( d@xmath34 ) and two 2p o orbitals ( p@xmath38 and p@xmath39 ) with on - site energies @xmath40 and @xmath41 , respectively . in this model
, an electron can hop from a d@xmath34 orbital in @xmath42 ( @xmath43 ) direction to a p@xmath39 ( p@xmath38 ) oxygen orbital and vice versa with the hopping integral @xmath44 ( @xmath45 ) .
additionally , due to strong hybridization of the oxygen 2p orbitals , electrons can hop between neighboring oxygen orbitals .
this leads to a hamiltonian of the form @xmath46 where @xmath47 are the creation operators for the above mentioned orbitals .
care has to be taken of the different signs of the hopping integrals due to the phase of the orbital wave - functions indicated by the tildes ( see fig .
[ fig : hopping1 ] ) . + to integrate out the high - energy degrees of freedom and thus , to reduce our model to one band , we construct an effective hamiltonian only living at the ru sites,@xcite @xmath48 where the sums run over all oxygen orbitals on all sites .
this leads to a simple hopping hamiltonian , @xmath49 where @xmath50 creates an electron at ru site @xmath51 with spin @xmath52 , @xmath53 denotes nearest neighbors ( nn ) and @xmath54 next - nearest neighbors ( nnn ) .
the hopping integrals in this effective hamiltonian then read to lowest order @xmath55 with @xmath56 . plane with rotated oxygen octahedra leading to a doubling of the unit cell .
the two distinct lattice sites are denoted as @xmath57 and @xmath58 .
the inversion symmetry of the bonds between ru - ions is broken leading to a staggered spin - dependent nearest - neighbor hopping .
the next - nearest - neighbor hopping is still isotropic and the same for both lattice sites as is depicted by the different , equivalent hopping - paths ( bold solid , dashed lines for @xmath57 to @xmath57 and dotted line for @xmath58 to @xmath58 ) . ]
if we additionally take the rotated oxygen - octahedra into account , we obtain a bipartite lattice leading to a @xmath59 larger unit cell separating a and b sublattices ( see fig . [
fig : rotated ] ) .
a consequence is that the formerly symmetry - forbidden hopping in @xmath42- ( @xmath43- ) direction via p@xmath38 ( p@xmath39 ) orbitals is now possible with matrix element @xmath60 as is depicted in fig .
[ fig : hopping2 ] .
thus , we include spin - orbit coupling at the oxygen site , @xmath61 , which mixes the @xmath62 and @xmath63 orbitals .
we therefore need to change to eigenfunctions of the spin - orbit coupling , @xmath64 , with @xmath65 , @xmath66 the spin index . + as an example , we write the total hamiltonian for the sublattice a for the @xmath42-direction , @xmath67 where @xmath68 creates an electron at the oxygen site @xmath69 in the @xmath64 state with spin @xmath52 and @xmath70 applying perturbation theory in the from of eq . to this hamiltonian to construct an effective model , we find for the hopping integral from a site @xmath57 to a site @xmath58 in the positive @xmath42-direction @xmath71 hence , we have a total hamiltonian @xmath72 with a nn - and nnn - hopping hamiltonian @xmath73 and a staggered spin - dependent hopping with the form of a staggered soc of rashba - type , @xmath74.\end{aligned}\ ] ] note that the nnn - hopping integrals , even though renormalized , do not become anisotropic , which can be deduced from geometrical considerations as indicated in fig . [
fig : rotated ] .
+ octahedra .
the sign change in the hopping integrals is due to the phase of the wannier functions and holds in first order . ] the bipartite lattice introduces a wave - vector @xmath75 with which the total hamiltonian in momentum space reads @xmath76 where @xmath77 . here
, @xmath78 are the hopping energies for nn- and nnn - hopping and @xmath79 is the form factor for the spin - orbit coupling .
the prime in the @xmath80- summation indicates that the summation runs only over the reduced first brillouin zone .
as this restriction will hold for all subsequent @xmath80-sums we will omit the prime in the following .
+ the staggered soc hybridizes states with @xmath81 and @xmath82 .
it is now convenient to introduce pauli matrices @xmath83 in momentum space @xmath84 such that we can write the hamiltonian as @xmath85 with @xmath86 diagonalizing this hamiltonian yields two ( spin - degenerate ) bands , @xmath87 where @xmath88 .
the first brillouin zone is folded back , as can be seen in fig .
[ fig : fs]a , where the fermi surface is plotted .
this is in accordance with the doubling of the unit cell introduced by the rotated oxygen octahedra . before adding an on - site interaction to the hamiltonian , we first want to examine the effect of an applied magnetic field . from the form of eq .
we first see that a magnetic field in @xmath89-direction is a mere spin - dependent shift of the chemical potential .
the response of the system is thus a simple polarization .
however , for in - plane fields the staggered spin - orbit coupling introduces a coupling of homogenous magnetic fields to a staggered magnetization , i.e. a commensurate sdw .
this means that the static spin - susceptibility has a component with wave - vector @xmath90 .
+ to see this , we add a zeeman term of the form @xmath91\ ] ] with the spin operators @xmath92 @xmath93 the bohr magneton and the land factor @xmath35 .
this corresponds to a homogenous and a staggered magnetic field , in accordance with the structure of the hamiltonian given in eq .
( [ eq : ham0 ] ) and can , thus , be written as @xmath94 here , we have introduced @xmath95 where @xmath96 .
it is now straightforward to calculate the magnetic response of the system to an applied field by using the thermodynamic relation @xmath97 where @xmath98 is the homogenous magnetization pointing in the @xmath51-direction while @xmath99 corresponds to a staggered magnetization .
using together with , we find for the case of a homogeneous field in @xmath42-direction a finite staggered magnetization in @xmath43-direction , @xmath100 in the above equation , @xmath101 are the four energy bands in a homogenous field in @xmath42-direction and @xmath102 is the fermi distribution function .
+ if we introduce an on - site interaction term to the hamiltonian , @xmath103 which we want to treat within mean - field theory , we first have to choose an appropriate spin - quantization axis ( indicated by the superscript @xmath104 ) . for simplicity
, we only consider the two cases of an applied magnetic field in @xmath89- and in @xmath42-direction .
since a field perpendicular to the plane does not couple to any staggered magnetization , the first case is straightforward .
the quantization axis is the @xmath89-axis and we write @xmath105 .\ ] ] since we do not expect large fluctuations in the charge density , the first term is a constant and , thus , the interaction can be written as @xmath106 with @xmath107 . applying mean - field theory to this expression and changing to momentum space yields @xmath108 with @xmath109 independent of site @xmath51 .
therefore , this leads to an additional term in ( [ eq : ham0 ] ) @xmath110 the total hamiltonian now reads @xmath111 with the effective magnetic field @xmath112 for simplicity .
this hamiltonian has four eigenenergies @xmath113 and via the grand - canonical potential per lattice site , @xmath114 + u ( m^{z})^2,\ ] ] the self - consistency equations can be derived , @xmath115 -plane : due to the staggered spin - orbit coupling , the magnetization is also staggered with respect to the sublattice sites a and b with order parameter @xmath116 and @xmath11 for the total moment and canting angle , respectively . ] .
right figure : density of states in the absence of an external field showing the van hove singularity at @xmath117 ( dashed line ) . ]
the case of a field in @xmath42-direction is slightly more involved since we expect an additional response in form of a staggered magnetization in @xmath43-direction leading to a canted magnetization in @xmath42-direction .
therefore , we define an angle @xmath11 for the canting angle and denote with @xmath116 the total moment ( see figure [ fig : canting ] ) . the quantization axis , @xmath118 and @xmath119 for the spin on lattice sites @xmath57 and @xmath58 , respectively , in eq .
should therefore be perpendicular to @xmath89 and tilted away from the @xmath42-direction by the angle @xmath11 .
we start from ( [ eq : spinspin ] ) and decouple to find @xmath120 changing again to momentum space , this results in an additional term in the total hamiltonian @xmath121 which then becomes @xmath122 with @xmath123 and @xmath124 the eigenvalues read @xmath125 again , the self - consistency equations can be deduced from the grand - canonical potential and are @xmath126 with @xmath127 from eqs .
( [ eq : sc - theta ] ) and ( [ eq : xi - theta ] ) it follows trivially that there is only a canted magnetization if there is a finite soc .
for an applied field in @xmath89-direction for a temperature of @xmath128 .
inset : the free energy @xmath129 for @xmath130 as a function of magnetization between @xmath131 and @xmath132 to emphasize the first - order character of the metamagnetic transition . ] since a metamagnetic transition is expected to occur close to magnetic instabilities , we investigate their occurrence from eq .
( [ eq : zmag ] ) .
the linearized self - consistent equation yields the condition @xmath133 for the occurrence of a ferromagnetic instability , the familiar stoner criterion .
this is not surprising since , apart from the folding of the brillouin zone , the staggered spin - orbit coupling in the case of a magnetic field in @xmath89-direction only leads to a renormalization of the nearest - neighbor hopping ( c.f .
( [ eq : xi - z ] ) ) .
the critical interaction strength for a ferromagnetic instability to occur as a function of the electron density , @xmath134 , is shown in the left part of fig .
[ fig : uc ] ( solid line ) .
it drops significantly close to a density of @xmath135 .
this corresponds to @xmath136 where the fermi surface hits the van hove points located at @xmath137 and @xmath138 thus leading to a diverging density of states .
this divergence is shown in the right part of fig .
[ fig : uc ] , where the density of states in zero field is plotted .
+ for a further analysis of the metamagnetic transition with applied field in @xmath89-direction , we fix the density of electrons slightly below @xmath139 , @xmath140 , and choose an interaction strength @xmath141 close to the critical one obtained from the linearized self - consistency equation ( see fig . [ fig : uc ] ) . here and in following numerical calculations
, we keep the spin - orbit coupling strength at @xmath142 .
the magnetization curve obtained is shown in fig .
[ fig : z - meta ] . to emphasize the first - order nature of the transition at @xmath128 ,
the inset shows the free energy at the critical magnetic field as a function of magnetization @xmath143 for values between @xmath144 and @xmath132 .
+ ( solid line ) and staggered magnetization @xmath145 ( dashed line ) for @xmath140 for an applied field in @xmath42-direction for a temperature of @xmath128 and @xmath146 .
for this temperature , the metamagnetic transition is clearly of first order . ] for the case of the field applied in @xmath42-direction , we can again explore the occurrence of a magnetic instability by linearizing the self - consistency equation , leading to @xmath147^{3/2 } } \big ] = 0.\ ] ] to further analyze this , it is useful to change to different order parameters , from ( @xmath148 ) to ( @xmath149 ) with @xmath150 the uniform magnetization in @xmath42-direction and @xmath151 the staggered component in @xmath43-direction , respectively . writing the self - consistency equations in these new parameters , @xmath152 and linearizing this system of equations , @xmath153 leads to two different possible magnetic instabilities , @xmath154 the first one is the same as eq . and corresponds to a ferromagnetic instability .
the second equation corresponds to a sdw instability occurring due to the near nesting of the fermi surfaces .
note , however , that in both cases , the magnetization will have a uniform as well as a staggered component .
+ the solutions of these equations as functions of the electron filling , @xmath134 , are plotted in fig .
[ fig : uc ] .
we see that the critical interaction strength for the spin - density wave instability is generally below the one for a ferromagnetic instability , but shoots up when approaching the critical filling .
+ we can now interpret the linearized equation as having a ferromagnetic and a spin - density wave contribution .
proximity to a sdw instability additionally lowers the critical interaction strength .
therefore , the metamagnetic transition could occur at a lower field then in the @xmath89-direction case , especially close to a sdw instability .
+ the magnetization due to a magnetic field applied in plane , as well as the amplitude of the staggered magnetization , are plotted in fig .
[ fig : xy - meta ] .
again , we find a first order transition for lower temperatures while the transition changes to a crossover upon increasing temperature .
note that the sign of @xmath151 depends on the sign of the spin - orbit coupling constant @xmath155 .
there is no degeneracy in the state obtained which could lead to domain formation .
+ without an applied field . due to the rotation of the o - octahedra
, the brillouin zone is folded back , denoted by the dashed lines .
( b ) and ( c ) : fermi surface just below and above the critical fields in @xmath89- and @xmath42-direction , respectively . for clarity ,
only a small section of the bz is shown indicated by the little square in ( a ) . for the case of a field applied in plane ( c ) ,
small gaps open close to the van hove points . ] comparing the two cases of fields applied in @xmath89- and @xmath42-direction in figs .
[ fig : z - meta ] and [ fig : xy - meta ] , respectively , we first see that the critical field for the latter is shifted to lower fields , even though the zero - field susceptibilities differ by less then a percent , @xmath156 .
this is due to the proximity of the system to a sdw instability as was pointed out in the previous subsection . +
this behavior is in qualitative agreement with the experimental phase diagram ( see the schematic phase diagram in fig . [ fig : schem - phase ] ) .
however , the difference of the in - plane and out - of - plane critical field is smaller in our model calculation than in the experiment .
our model only includes the staggered spin - orbit coupling entering through the oxygen displacement .
naturally , other spin - orbit coupling contributions , particularly from the ru - ions , would add to the anisotropy through an anisotropic @xmath35-tensor , likely with a larger polarizability in the basal plane than along the @xmath89-axis .
@xcite this is , however , beyond the scope of this study as a detailed analysis would require to include other bands .
+ second , a numerical study of the gibb s free energy shows that the temperature @xmath157 up to which the first - order transition persists , is higher in the case of the in - plane field . for our choice of parameters
we find @xmath158 while @xmath159 .
this anisotropy in the critical temperature is consistent with the trend in the experimental situation .
however , it does not reproduce the quantum critical endpoint . note , that the difference between @xmath160 and @xmath161 could not be explained simply by an anisotropic @xmath35-tensor . in principle , it may be possible to tune the model in such a way as to press the critical temperature for out - of - plane fields , @xmath162 to zero while still having a first - order transition at finite temperatures for in - plane fields .
also fluctuation effects are likely important in this context .
these features are , however , not essential to our discussion . + an additional important difference between in - plane and out - of - plane fields can be seen in fig .
[ fig : fs ] , where the fermi surfaces for both cases for fields below and above @xmath163 is shown : we see that the system undergoes a metamagnetic transition to prevent the majority - spin band from touching the van hove points . in ( c )
, we additionally see that the induced spin - density wave opens small gaps at the fermi level close to the van hove points .
this has important consequences for the appearance of a nematic phase as we will see in the next section . for a field applied in @xmath89-direction for a temperature of @xmath128 showing an intermediate nematic phase bounded by two first - order transitions .
here , @xmath164 and @xmath140 . ]
in this section , we explore the occurrence of a nematic phase in our model for the two cases of a magnetic field applied in @xmath89- and @xmath42-direction , respectively . for this purpose
, we introduce an additional interaction term @xcite @xmath165 with a coupling function @xmath166 only contributing for zero momentum transfer , i.e. for the forward scattering , which is the relevant interaction for a nematic phase to occur .
@xcite we then separate the coupling function @xmath167 and choose a d@xmath168 symmetric form for the form factors , @xmath169 .
this term can then lead to a nematic phase , reducing the symmetry from @xmath170 to @xmath171 . + introducing again a mean - field decoupling which is spin - independent we write for this interaction @xmath172 with @xmath173 since @xmath174 , but is isotropic in spin - space , we can deal with it by replacing @xmath175 while all the above formulae still hold with the additional self - consistency equation @xmath176[eq : nem - z ] for the @xmath89-direction case and @xmath177 for the @xmath42-direction case , respectively . + -direction .
while for low temperatures , the two consecutive transitions are of first - order ( solid line ) , they become second - order before the nematic phase disappears completely ( dashed line ) . above this temperature
, a metamagnetic crossover can still be seen . ] for sufficiently strong @xmath35 , we find a magnetization curve for fields applied in @xmath89-direction as is shown in fig .
[ fig : nem - z ] .
the two jumps in the magnetization border an intermediate phase with a finite value of the nematic order parameter @xmath178 .
the instability in this case is again driven mainly by electrons whose momenta lie close to the van hove points . to obtain an intermediate phase before a single metamagnetic jump removes all such electrons from the fermi surface ,
a critical scattering strength @xmath179 is necessary . above that , a nematic phase is entered at some magnetic field @xmath180 and left again at @xmath181 . +
the @xmath27-@xmath5-diagram shown in fig.[fig : phase - z ] shows first - order transitions for low temperatures up to @xmath182 , second - order transitions for higher temperatures until at @xmath183 the nematic phase disappears completely to make way for a metamagnetic cross - over ( not shown ) .
this behavior has already been observed in similar calculations .
@xcite + for the case of a field applied in @xmath42-direction , a very similar behavior is observed , however , with one important difference : since the induced spin - density wave already removes some weight from the fermi surface close to the van hove points , a larger forward scattering strength is required for the occurrence of a nematic phase , i.e. @xmath184 .
therefore , there is a range of @xmath35 , where there exists already a nematic phase for fields in @xmath89-direction , but only one metamagnetic jump is observed for in - plane fields .
this result is summarized in the phase diagram in fig .
[ fig : nemz - nemxy ] . as a function of the soc strength @xmath155 and the forward scattering strength @xmath35
, we find three regions .
in addition to the two obvious ones , where there is either no intermediate phase ( region i ) at all or one for fields applied in any direction ( iii ) , there is now a new region with a nematic phase only for fields applied in @xmath89-direction ( ii ) .
obviously , this region corresponds to the case of sr@xmath0ru@xmath1o@xmath2 . - and @xmath42-direction , respectively .
we can distinguish three different regions .
i : the forward scattering strength is too week to enter a nematic phase , no matter in what direction the magnetic field is applied .
only a single metamagnetic transition occurs .
ii : while there is a nematic phase for fields applied in @xmath89-direction , no such phase occurs for fields in plane .
this region corresponds to the case found in sr@xmath185ru@xmath9o@xmath10 .
iii : the forward scattering strength is strong enough such that a nematic phase will occur for fields in any direction . _
inset _ : schematic of the phase diagram showing that the nematic phase only occurs for fields close to the @xmath89-axis . ]
the range of the forward scattering strength for which in our calculation a nematic phase is only observed for fields in @xmath89-direction is not very large .
note , however , that the nematic phase only appears in a very narrow region ( @xmath186@xcite ) and thus , @xmath35 is only slightly bigger than @xmath187 .
this is illustrated in fig .
[ fig : g - h ] where the dependence of the critical fields on the forward scattering strength is shown .
the width of the nematic phase grows rather rapidly with increasing @xmath35 .
therefore , the actual size of the forward scattering strength might well lie in region ii of fig .
[ fig : nemz - nemxy ] .
+ obviously , not only the strength of the soc , but also the nesting properties of the spin - polarized fermi surfaces play an important role for the appearance of the anisotropy effect .
nesting properties are a factor of tuning our model to the vicinity of a sdw instability . to analyze the impact of enhanced sdw correlations , we consider two different electron densities and corresponding on - site interaction strengths @xmath141 with @xmath188const ( see crosses in the left part of fig.6 ) .
this allows us to examine the cases of two different proximities to a sdw instability with comparable strengths of fm correlations .
one important finding is that the stronger the sdw correlations the more pronounced the anisotropy effect and thus , the smaller the ratio @xmath189 becomes.this is depicted in fig .
14 . in fig .
6 we also show that the ferromagnetic and sdw instabilities can be competing for the chosen parameter range .
we fix our model parameters in a way to avoid the occurrence of a staggered magnetic moment for any value of the magnetic field along the z - axis , while the staggered moment is field induced for in - plane fields .
+ we should also comment on the strength of the soc @xmath155 that we expect for this system . to get an estimate of the on - site soc strength @xmath190 for p - electrons on the oxygen , we take the o@xmath191 vacuum value , @xmath192mev . a very crude estimate of the staggered soc coupling from eq
. then yields @xmath193 taking estimates for @xmath194 we find that for @xmath195 a value on order of a percent of @xmath196 seems reasonable .
@xcite comparing figs .
[ fig : nemz - nemxy ] and [ fig : g - h ] , this would allow for a nematic phase with a width of @xmath197 , in agreement with experiment . for a more reliable estimate of @xmath155 , dft calculations should be performed .
+ finally , some remarks to the nematic phase are in order .
as was already mentioned , the nematic phase introduced here is the same as discussed by other authors .
@xcite as shown by these authors , the nematic phase could account for several experimentally observed phenomena , like the anomalous resistivity , or the non - fermi - liquid behavior of the susceptibility and the specific heat coefficient .
+ -direction depending on the forward scattering strength @xmath35 . below a critical strength @xmath198
, no nematic phase is entered , but the system undergoes a single metamagnetic transition connected to the proximity to the ferromagnetic instability as discussed above . above @xmath198 ,
the system enters a nematic phase whose region grows with increasing @xmath35 . here , @xmath128 .
] to conclude , we showed that the rotated oxygen octahedra lead to a staggered hopping that can be described with the help of a ( staggered ) spin - orbit coupling of rashba - type .
this introduces an anisotropy of the response to a magnetic field , namely an induced spin - density wave for the case of in - plane fields .
this staggered magnetization could be observed in neutron scattering experiments . to our knowledge
this kind of experiment has not been performed so far .
the additional magnetization has , first , the effect that the critical field for a metamagnetic transition is shifted to lower values for in - plane fields .
also , the critical temperature @xmath199 up to which the transition is first order is higher for fields in the @xmath200-plane .
last and most important , the spin - density wave opens gaps at the fermi level that lead to an anisotropy for the appearance of a nematic phase . additionally considering spin - orbit coupling effects of
the ru orbital would account for the full anisotropy of @xmath3 ( @xmath35-tensor anisotropy ) .
therefore , the present work allows for a picture that is qualitatively consistent with experimental observations including the anisotropies in @xmath3 , @xmath199 and the appearance of a nematic phase .
+ note added : while preparing for submission we noticed the recent paper @xcite which studies the influence of spin - orbit coupling and the doubling of the unit cell on the nematic phase of sr@xmath185ru@xmath9o@xmath10 .
we are grateful to h. adachi , b. binz , f. hassler , f. loder , a. mackenzie , y. maeno , t.m .
rice , a. regg , m. ossadnik and a. thomann for helpful discussions .
this work was financially supported by swiss nationalfonds and by the nccr manep .
-direction compared to the value for the @xmath89-direction for the densities @xmath140 ( solid line ) and @xmath201 ( dashed line ) , respectively .
the value for the on - site interaction is chosen as indicated in fig .
[ fig : uc ] .
the dependence on the strength of the spin - orbit coupling is more pronounced the closer the system is to a sdw instability . ]
|
ultra - clean crystals of sr@xmath0ru@xmath1o@xmath2 undergo a metamagnetic transition at low temperatures .
this transition shows a strong anisotropy in the applied field direction with the critical field @xmath3 ranging from @xmath4 t for @xmath5 perpendicular to @xmath6 to @xmath7 t for @xmath8 .
in addition , studies on ultra - pure samples revealed a bifurcation of the metamagnetic line for fields in @xmath6-direction and it has been argued that a nematic phase emerges between the magnetization jumps . the aim of this study is to explain the field - direction anisotropy of these phenomena .
based on a microscopic tight - binding model , we introduce the metamagnetic transition by means of a van hove singularity scenario .
we show that the rotation of the o - octahedra around the @xmath6-axis observed in this material introduces a staggered spin - orbit coupling within the planes and naturally leads to an anisotropy in the low - temperature behavior around the metamagnetic transition . in particular , the low - temperature ( nematic ) phase is affected .
we show that uniform in - plane magnetic fields induce a ( commensurate ) staggered magnetic moment component which can suppress the low - temperature phase .
in contrast , the response to fields along the c - axis remains unaffected and thus , also the corresponding low - temperature phase . as a concrete example
, we choose a nematic pomeranchuk instability for the low - temperature phase .
an experimentally testable prediction of this work is the occurrence of a staggered magnetic moment in response to a uniform magnetic field perpendicular to the c - axis , which should be accessible by neutron scattering .
|
recently , in several studies@xcite of heavy - ion - induced nuclear multifragmentation it was concluded that certain characteristics of the intermediate - mass fragment ( imf ) data are suggestive of statistical independence of multiple imf production.@xcite concepts of a binomial or a poissonian reducibility@xcite have been suggested to express the particular observed independence in mathematical terms .
the concept of statistical independence or reducibility presumes that certain characteristics of multi - imf events are reflections of the single - imf production process .
for example , the concept of a binomial reducibility@xcite presumes that the probability of multiple imf production is a reflection of the single - imf production probability - the two quantities of interest being connected via the expression for a binomial distribution : @xmath2 where @xmath3 is the multiple - imf production probability and the parameters @xmath4 and @xmath5 are the number of ( binomial ) tries and the probability for success in any of these trials , respectively .
although so far , no theoretical model has been proposed that would explain a binomial reducibility , one would reasonably expect that a successful model would at the same time link other characteristics of multi - imf events to the characteristics of single - imf production . with these expectations in mind ,
the present paper investigates , to what extent the empirical systematic of multiple imf production can be derived from empirical characteristics of single - imf production and , specifically , from the single - imf transverse - energy spectrum .
it is assumed in this study that the hypothesis of an independent imf production holds and that , accordingly , the total transverse - energy spectra of multiple imfs can be synthesized from the transverse - energy spectrum of a single imf .
section 2 presents such a synthetic procedure , based on multiple folding of a schematic single - imf spectrum .
section 3 discusses results obtained when empirical single - imf spectra are substituted for the schematic distributions used in section 2 , and section 4 discusses effects of an inclusion of transverse energies of light charged particles in the definition of the total transverse energy .
an alternative synthesis method based on fragment swapping across different events of different imf multiplicity is discussed in section 5 .
section 6 summarizes the findings of the present study .
in the following , the total transverse energy of charged reaction products is defined as @xmath7 where the summation extends over all particles with atomic numbers @xmath8 . in the present section , only imfs with @xmath9
are included in the definition of @xmath0 .
effects of an inclusion of light charged particles are discussed further below , in section 4.the following empirical characteristics of multi - imf events represent the rationale behind the synthesis procedure presented in this section : 1 .
the @xmath6 distributions of individual imfs are approximately exponential , with inverse logarithmic - slope parameters @xmath10 of the order of 30 to 50 mev.@xcite as an example , the experimental transverse - energy spectrum is shown in fig . 1 , for single imfs from the bi+xe reaction@xcite at @xmath11 mev .
since @xmath6 is an angle - integrated quantity , the corresponding spectrum may differ from a maxwell - boltzmann shape .
superimposed on the data are an exponential distribution with @xmath12 mev ( dotted line ) and a two parameter gaussian fit ( solid line ) to the spectrum . while the exponential function deviates from the experimental spectrum both at high and low values of @xmath0 , it simplifies the following demonstration of qualitative features of total imf @xmath0 spectra .
+ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ fig .
intermediate - mass fragment transverse - energy spectrum of single - imf events .
the dotted line is an exponential distribution , @xmath13 , used as a first
order approximation of single
imf spectra in eq .
[ eq : expo ] .
the spectrum itself was used as a model distribution @xmath14 for a numerical evaluation of the integrals ( [ eq : foldingimfs ] ) and ( [ eq : foldinglcps ] ) with a monte carlo method .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 2 . to a good approximation ,
@xmath6 spectra of individual imfs are independent of the imf coincidence fold ( multiplicity ) @xmath15.@xcite although discernible , the dependence on @xmath15 is weak .
3 . in addition to imfs
, one also observes light charged particles ( lcps ) , which greatly outnumber imfs , for every @xmath15 . even for @xmath16 ,
the average multiplicity of lcps is of the order of 10 and , hence , quite significant .
4 . at low bombarding energies , e.g. ,
for @xmath17 mev , the imf contribution can dominate the total transverse energy @xmath0 , for all but the lowest multiplicities @xmath15.@xcite for larger bombarding energies , the relative imf contribution decreases but remains always significant . assuming a statistically independent imf production , the probability distribution @xmath18 for events with imf multiplicity @xmath15 can be expressed in the form of a product of @xmath15 individual single imf distributions @xmath19 @xmath20 in eq .
[ eq : independence ] , as in the following discussion , single
particle variables will be denoted by lower - case characters , while upper - case characters indicate multi particle variables or distributions .
for example , the symbol @xmath21 will be used to denote multi particle probability distributions , while the symbol @xmath22 indicates the corresponding single particle distribution .
similarly , the variable @xmath23 denotes the contribution of particle number @xmath24 to the total ( multi - particle ) transverse energy @xmath6 .
while the first part of eq .
[ eq : independence ] represents the statistical independence , the second expresses the equality of all individual energy distributions .
given the above form ( [ eq : independence ] ) of @xmath25 , the distribution @xmath26 of the total transverse energy can be expressed as @xmath27 where the integration is carried out over a hyperplane of a constant @xmath6 , as enforced by the @xmath28 function .
equation [ eq : generalfolding ] can be further simplified to : @xmath29 in the general case of an arbitrary individual single - particle distribution @xmath22 , the integration in eq .
[ eq : foldingimfs ] can be performed numerically , at least for reasonably low imf multiplicities .
when a simple exponential function is taken to approximate @xmath22 , this integral can be evaluated analytically for arbitrarily large imf multiplicities : @xmath30 @xmath31 in principle , the inverse slope parameter @xmath32 could depend on the imf multiplicity @xmath15 , but the experimental dependence on @xmath15 is known to be weak.@xcite according to eq .
[ eq : poiss ] , to the extent that the single - imf transverse - energy spectra can be approximated by exponentials , the distribution of the composite observable @xmath6 is of poissonian form for each imf coincidence order @xmath15 . in order to calculate quantitatively the two - dimensional joint probability distribution @xmath33 for a given reaction
, one should combine the one - dimensional @xmath6 distributions ( [ eq : poiss ] ) for given @xmath15-values with the imf multiplicity distribution @xmath34 , measured for this reaction , i.e. , @xmath35 typical experimental @xmath34 distributions are depicted in fig . 2 . for the experimental data discussed here
, they can be approximated well by gaussians , @xmath36 ( solid line in the figure ) .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ fig . 2
. intermediate - mass fragment yield distribution of @xmath15fold imf events , for @xmath37 .
the solid line is a gaussian distribution , @xmath38 $ ] .
this yield distribution was used for a numerical evaluation of the integrals ( [ eq : foldingimfs ] ) and ( [ eq : foldinglcps ] ) with a monte carlo method . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the joint distribution in eq .
[ eq : weightedpoiss ] allows one to synthesize conditional imf multiplicity distributions , gated with different values of @xmath0 , and to calculate the binomial parameters @xmath4 and @xmath5 for these distributions for further analysis along the lines adopted in a series of recent papers.@xcite specifically , one calculates @xmath39 @xmath40 @xmath41 @xmath42 _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ fig .
lower panel : the distributions ( [ eq : poiss ] ) are plotted up to the order of @xmath43 .
note the qualitative similarity of the distributions ( [ eq : poiss ] ) and the @xmath0 distributions due to multifragmentation , plotted in the subsequent figures .
upper panel : the values of parameters @xmath5 and @xmath4 calculated with equations ( [ eq : p ] ) and ( [ eq : m ] ) and distributions ( [ eq : poiss ] ) , with sharp
cutoff weights @xmath44 , for n up to @xmath45 , and @xmath46 for @xmath47 .
three example values of @xmath48 are indicated in the figure .
note the severity of the truncation effect , which causes the calculated @xmath5 and @xmath4 to deviate from their limiting values , zero and infinity , respectively .
the limits are reached only for @xmath45 approaching infinity .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ it is the use of the constrained empirical distribution @xmath34 that leads to a reduction of the variances of the conditional multiplicity distributions below the poissonian limit of @xmath49 and , hence , ensures positive values for the parameters @xmath4 and @xmath5 .
this is illustrated in fig . 3 , where a schematic sharp cutoff distribution of @xmath34 serves to demonstrate the importance of truncation .
note the large magnitude of the truncation effect , which causes the calculated parameters @xmath5 and @xmath4 to deviate from their limiting values , zero and infinity , respectively .
these limits are reached only when the number of folds approaches infinity .
significant values for both @xmath5 and @xmath4 have been generated within the physically relevant range of @xmath0 by restricting the number of folds to @xmath50 , i.e. , a number of coincidence folds typically measured in an experiment .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ fig . 4 .
lower panel : the @xmath0 distributions ( [ eq : weightedpoiss ] ) weighted with the experimental imf yields @xmath34 are compared with experimental imf only transverse - energy distributions .
the upper part of the figure shows the respective @xmath5 and @xmath4 parameters as calculated with eqs .
( [ eq : p ] ) and ( [ eq : m ] ) , respectively . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the synthetic distributions are compared in fig . 4 with experimental imf transverse - energy distributions . in the calculations ,
the imf yields @xmath34 were taken from experiment.@xcite the inverse slope parameter @xmath51 mev of the imf spectra was adjusted to approximately match the singles ( @xmath52 ) spectrum up to about 200 mev ( see fig . 1 ) .
as seen in this figure , already a synthesis using an exponential singles spectrum reproduces qualitatively the gross features of the experimental @xmath6 distributions .
the lack of a quantitative agreement with experimental data is not surprising , since the used schematic exponential single
particle distribution ( [ eq : expo ] ) is only a very crude approximation of the actual single particle spectrum . as seen in fig . 1
, the tail of the exponential function does not match the @xmath52 spectrum for @xmath53 mev .
nevertheles s , the overall features of the experimental @xmath6 distributions are reproduced qualitatively by those calculated with eq .
[ eq : weightedpoiss ] . the upper part of fig .
4 presents a comparison between the parameters @xmath5 and @xmath4 , extracted from experimental data , with the ones calculated from eqs .
[ eq : meann ] and [ eq : sigman ] , using the distributions ( [ eq : weightedpoiss ] ) displayed in fig . 2 . the predicted values of @xmath5 and
@xmath4 match the experiment remarkably well , even though the calculated @xmath6 distributions in the lower panel deviate from the experimental ones .
this points towards the need for a better description of experimental single
fragment @xmath6 distributions @xmath14 .
more realistic shapes of these single particle distributions will be considered next .
a more realistic rendering of the shape of the single imf distribution than achieved by an exponential is provided by a two parameter gaussian fit of the kind depicted by the solid line in fig . 1 .
such a gaussian spectrum was used in a numerical evaluation of the integrals ( [ eq : foldingimfs ] ) , employing a monte - carlo sampling method results of which are displayed in fig .
5 . as seen in fig . 5 ,
a remarkable quantitative agreement with experimental data is achieved .
this time , also the parameters @xmath5 and @xmath4 have been reproduced .
it is worth emphasizing , that only four parameters , two shape parameters of the single - particle spectrum ( cf .
fig . 1 ) and two parameters describing the imf yield distributions ( cf . fig .
2 ) ) are sufficient to reproduce the shapes of the multi imf @xmath0 distributions , and the @xmath0-dependencies of the parameters @xmath5 and @xmath4 .
in order to better represent the experimental situation , the contribution by light charged particles to the total kinetic energy has to be considered .
this can be achieved easily by extending the concept of a statistically independent fragment production to the lcp emission . in this case , @xmath54 the index 0 now denotes the lcp contribution to the total @xmath6 .
since there are many lcps contributing independently ( even for the case of no imfs , @xmath16 ) , according to the central limit theorem , the contribution of all lcp s , taken together , is approximately a gaussian @xmath55},\ ] ] where the parameters @xmath56 and @xmath57 may depend on the imf multiplicity @xmath15 . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ fig . 5 .
lower panel : imf transverse- energy distributions ( solid lines ) , predicted by folding a gaussian form of the @xmath52 imf distribution multiple times with itself ( without the lcp contribution ) , according to formula ( [ eq : foldingimfs ] ) .
the resulting @xmath0 spectra were weighted with the gaussian imf yields @xmath34 from fig . 2 .
the symbols connected with dotted lines represent experimental data .
upper panel : @xmath5 and @xmath4 parameters extracted from the experimental data ( symbols connected with lines ) , compared with the ones extracted from the set of folded @xmath0 distributions .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ wieloch et .
al.@xcite have considered the particular parameterization of @xmath58 and @xmath59 .
the present paper employs values of @xmath56 and @xmath57 taken from experiment,@xcite namely @xmath60 for @xmath61 and @xmath62 for @xmath63 . with this choice of parameter values @xmath56 and
@xmath57 , the integral ( [ eq : foldinglcps ] ) was evaluated numerically in the same way as before .
the results are displayed in fig .
6 , showing satisfactory agreement with experimental data .
the parameters @xmath5 and @xmath4 are also rather well reproduced by the calculation , except for the deviations from experimental data at @xmath64 mev .
the deviations are attributed to an imperfect representation of lcp spectra by gaussians , and in particular to the intense contributions of peripheral events to the lcp spectra at low @xmath0 .
it should be noted , that low@xmath0 data are usually considered less important in the context of multifragmentation .
as an alternative approach to synthesizing multi - imf events , the experimental data@xcite have been subjected to `` event mixing '' , a procedure where both imfs and lcps are randomly exchanged between different events . for every event ,
the number of lcps , @xmath65 , and the number of imfs , @xmath15 , were preserved , but not the identities of these particles . both imfs and lcps were swapped among events of all multiplicities ( regarded as a common `` fragment pool '' ) , in order to ensure a complete `` loss of memory '' in the synthetic mixed
event data set .
the above random particle exchange between events ensured that multi - particle correlations were erased from experimental data , while preserving approximately the shape of the single - particle distributions . in this fashion , a statistical independence among particles
was strictly imposed on experimental data , even if some degree of multi - particle correlations had been present in the original data set .
this procedure ensured that the assumptions underlying eq .
[ eq : foldinglcps ] were strictly fulfilled by the data .
it should be stressed , that event mixing does not represent any additional assumption brought into the problem .
rather , it is a convenient way to evaluate the integral ( [ eq : foldinglcps ] ) , where all single particle probability distributions were sampled directly from the experimental data , without recourse to a particular parameterization .
subsequently , the synthetic events were subjected to the same analysis procedure as the original data .
the results of this mixed - event analysis are compared with the original data in figs .
7 and 8 . very good agreement between both sets of results is observed from these figures , both with regard to the @xmath6 distributions and the extracted parameters @xmath5 and @xmath4 .
this result confirms again that the concept of a statistically independent imf production is a valid one . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ fig .
the @xmath5 and @xmath4 parameters extracted from the experimental data set ( symbols connected with solid line ) and from the same set subjected to random particle swapping ( dotted lines ) , plotted in an arrhenius like representation as functions of @xmath66 .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
accepting the validity of the findings of earlier papers@xcite regarding the statistical independence , or reducibility , of multiple intermediate - mass fragment production , one finds that important characteristics of multi - imf events can be derived already from the experimental shapes of single - imf transverse energy distributions .
the characteristics of the synthetic events generated from these distributions show remarkable resemblance to those of the experimental multi - imf events .
the latter characteristics include the binomial reducibility and an arrhenius - like scaling discovered in recent studies of nuclear multifragmentation.@xcite although the findings of the present study do not directly favor one imf production scenario over another , they do so indirectly .
they draw attention to the crucial importance of an understanding of the shapes of the single - imf @xmath0 distributions . in the present analysis , the apparent binomial reducibility and the apparent arrhenius - like
scaling are secondary phenomena , both reflecting the `` input '' shape of the elementary @xmath0 distribution .
hence , a valid interpretation of the significance of such reducibility and scaling in terms of a particular imf production scenario requires an independent explanation of the underlying single - imf @xmath0 distribution . a purely thermal scenario , which could potentially justify arrhenius - like correlations between emission probability and temperature of the system , makes quite definite predictions regarding the elementary @xmath0 distributions , which should be very similar to those of the lighter particles emitted from the same sources .
these predictions disagree with the experimental imf spectra , which are significantly harder than the light - particle @xmath0 spectra .
this observation justifies some skepticism towards an interpretation@xcite of arrhenius - like plots in terms of thermal scaling .
the bi+xe experiment at e / a = 28 mev was conducted in collaboration with s.p.baldwin , b. m. quednau , l.g . sobotka , j. barreto , r. j. charity , l.gallamore , d.g .
sarantites , d. w. stracener , r. t. de souza , and b.lott . their contribution is highly appreciated .
this work was supported by the u. s. department of energy grant no .
de fg0288er40414 .
99 present address : x - ray instrumentation associates , 2513 charleston road , suite 207 , mountain view , ca 94043 , usa l.g .
moretto , r. ghetti , l. phair , k. tso , g.j .
wozniak , phys .
287 , 249 ( 1997 ) , and references therein .
j. tke , et al .
lett . * 77 * , 3514 ( 1996 ) .
j. tke et .
al . , _ advances in nuclear dynamics 2 , _ proceedings of 12th winter workshop on nuclear dynamics , february 3 - 10 , 1996 , snowbird , utah . edited by w. bauer and g.d.westfall , plenum press 1996 .
a.wieloch et .
al . , z.phys .
a359 , 345 ( 1997 ) .
a.wieloch et .
al . , phys .
b*432 * , 29 ( 1998 ) .
w. skulski et .
, report doe / er/40414 - 8 , page 75 and 86 ( 1995 ) .
|
it is studied to what extent the characteristics of multi - imf ( intermediate - mass fragments ) events can be derived from the properties of the observed single - imf transverse - energy spectra .
it is found that the spectra of total transverse imf energy ( @xmath0 ) in multi - imf events are well represented by `` synthetic '' spectra obtained by a multiple folding of the single - imf transverse energy spectrum .
further , it is shown that , using the experimental imf multiplicity distribution in the folding procedure , it is possible to reproduce the observed trends in the imf multiplicity distributions for fixed values of the total transverse energy @xmath0 .
accordingly , the `` synthetic '' multiplicity distributions show a binomial reducibility and an arrhenius - like scaling similar to that reported in the recent literature .
consistent with the concept of statistical independence of multiple imf production , similar results are obtained when the above folding - type synthesis is replaced with one based on mixing of events with different imf multiplicities .
for statistically independent imf emission , the observed binomial reducibility and arrhenius - like scaling are shown to be merely reflections of the shape of the single - imf transverse - energy spectrum .
hence , a valid interpretation of these imf distributions in terms of a particular production scenario has to explain independently the observed shape of the single - imf @xmath0 spectrum .
# 1@xmath1#1 ( pacs 25.70lm , 24.10.pa , 25.70.mn , 25.70.pq )
|
borromean states are most easily found for potentials that can almost bind a two - body system .
the relative motion in two - body systems is described with the wave function given as a product of radial , @xmath2 , and angular , @xmath3 , parts , where @xmath4 are the polar relative coordinates , and @xmath5 is the 2d angular momentum quantum number .
the radial schrdinger equation is @xmath6 where the bounded potential , @xmath7 , of cylindrical symmetry is assumed to decrease faster than @xmath8 , @xmath9 , for large @xmath10 , or in practice treated as zero outside a finite radius @xmath11 .
the two - body energy is @xmath12 , @xmath13 , is a dimensionless strength parameter , and @xmath14 is the mass of one particle .
the doubly degenerate antisymmetric ground state for spin polarized fermions has @xmath15 , since @xmath16 describes a symmetric total wave function .
therefore , for two fermions we shall only consider @xmath17 and omit the related index .
the regular zero energy solution , @xmath18 , to eq
. obeys @xcite @xmath19 the large - distance asymptotic of this @xmath20 solution is uniquely determined by the length parameter , @xmath21 , defined such that @xmath18 asymptotically approaches @xmath22 @xcite , where @xmath23 the critical value , @xmath24 , for two - body binding is reached when @xmath25 where the zero - energy solution crosses zero at @xmath26 , see methods .
then @xmath24 is defined as the smallest value of @xmath27 where the denominator in eq .
the limit @xmath28 corresponds to @xmath29 , since binding of two identical fermions in 2d requires a finite attraction .
to address weakly bound two - fermion systems we use the stochastic variational method @xcite to calculate ground state energies for different values of @xmath27 . to illustrate the generic nature of our findings
, we choose three qualitatively different potentials @xmath30 .borromean binding energies . numerically calculated two - body thresholds for binding , three - body properties at the thresholds and estimates for three - body thresholds for binding .
ground and excited states have angular momentum @xmath31 and @xmath32 .
lengths and energies are in units of @xmath33 and @xmath34 , respectively . [ cols="^,^,^,^,^,^,^",options="header " , ] the first has a simple attractive pocket , the second has a repulsive barrier outside the attraction , and the third has a repulsive core .
the calculated values of @xmath35 are given in tab .
the two - body bound states with @xmath36 are universal , independent of potential , i.e. the wave function is a modified bessel function @xmath37 @xmath38 , corresponding to @xmath39 with essentially only non - zero probability in the classically forbidden region of zero potential . using the wave function , @xmath40 , in all space we find the mean square radius for small @xmath41 , i.e. @xmath42 where we used the potential range parameter , @xmath33 , to get a dimensionless argument of the logarithm . ) .
the points are obtained numerically by decreasing the attraction @xmath27 through the variation @xmath43 .
the curves are produced by fits with a length parameter , @xmath44 , in eq .
( [ as - r ] ) different from @xmath33 , i.e. we have fitted the numerical results to the functional form on the left - hand side of eq .
( [ as - r ] ) .
this incorporates next - to - leading order terms proportional to @xmath45^{-1}$ ] .
the limit for @xmath46 is @xmath47 .
the parameters of the fits are @xmath48 for @xmath49 , @xmath50 for @xmath51 and @xmath52 for @xmath53 . ] to confirm our understanding through eq
. is numerically challenging for some potentials . in fig .
[ figrms ] we see that the limit of @xmath47 only is reached for very small @xmath12 , although much faster for @xmath49 and @xmath51 than for @xmath53 .
the different rates of convergence illustrate how small an energy is necessary to reach the universality limit for potentials of varying structure .
the similarity for @xmath49 and @xmath51 arises because the repulsive core for @xmath51 is ineffective for the @xmath54-waves , while their attractions are similar . on the other hand ,
the external barrier for @xmath53 delays tunneling into the universal region of the classically forbidden space and beyond .
the radius increases with decreasing @xmath41 as larger distances are populated .
however , the universal limit will eventually be reached for sufficiently small @xmath12 .
these properties are markedly different for two bosons in @xmath55 where the symmetric wave function with @xmath16 is well described by @xmath56 at comparably small binding energies @xcite .
these observations are of crucial importance in understanding the universal two - body bound states that arise near the window for borromean binding .
the two - fermion binding requirement of a finite potential depth in @xmath55 resembles that of two bosons in @xmath57 .
this strongly suggests that borromean three - fermion systems are possible in @xmath55 even for purely attractive potentials .
also , the efimov effect can not be strictly ruled out for three spin - polarized fermions in @xmath55 @xcite .
in contrast , we know that the efimov effect is absent for three bosons in 2d , and borromean states only occur for potentials with positive net volume , @xmath58 , and regions with substantial repulsion @xcite . in 3d , the coordinate - space adiabatic hyperspherical expansion method proved to be very efficient for short - range interactions due to small large - distance couplings @xcite .
this efficiency is highlighted in the precise description of the efimov effect by use of only the lowest adiabatic potential @xcite .
we shall therefore here employ the 2d hyperspherical formalism @xcite for three spin - polarized fermions .
let @xmath59 be the coordinate of the @xmath60th particle .
one set of the relative jacobi coordinates is @xmath61 and @xmath62 , while the other sets are obtained by cyclic permutation of @xmath63 .
the hyperspherical coordinates are given by @xmath64 and the three angles for each jacobi set , @xmath65 , where @xmath66 , and @xmath67 and @xmath68 define directions of the coordinates @xmath69 .
the kinetic energy operator has the form @xmath70 where the hyperangular part is @xmath71 the eigenfunctions corresponding to the dependence on @xmath67 and @xmath68 are @xmath72 , where @xmath73 and @xmath74 are integers .
the sum , @xmath75 , is a conserved quantum number which labels the solutions by @xmath76 .
the total wave function , @xmath77 , is expanded in a complete set of hyperangular functions , @xmath78 , for each @xmath79 chosen as eigenfunctions of the hyperangular part of the hamiltonian , @xmath80 .
thus , @xmath81 where @xmath82 is the two - body potential , and the functions @xmath83 satisfy a system of coupled equations .
we omit the index @xmath84 and search only for the angular solution , @xmath85 , of lowest energy for a given @xmath86 .
we use the faddeev decomposition and expansion on the eigenfunctions related to @xmath67 and @xmath68 , that is @xmath87 where @xmath88 and the index
@xmath73 only assumes odd values to ensure antisymmetry . and @xmath17 at short and long distance .
the inset shows a zoom of the behavior at large distance to illustrate the barrier in the @xmath16 channel .
note that only the vertical axis has been amplified in the inset .
the inset starts roughly at the distance @xmath89 .
for the potentials considered here the attractive pocket in the @xmath16 channel is considerably larger than in the @xmath17 channel .
we therefore have stronger binding in the @xmath16 channel . ]
we first solve to lowest order without any non - adiabatic and coupling terms , that is we find @xmath90 in the so - called exact adiabatic approximation @xcite , from @xmath91 using the lowest adiabatic potential produced by @xmath92 we then get a lower bound on the energy @xcite .
the behavior of @xmath92 for large @xmath79 is decisive for sufficiently weakly bound states provided the centrifugal barrier is small .
we find @xmath92 by solving eq . with the structure of @xmath85 in eq .
in general , the contributing configurations in angular space arise from the smallest @xmath93-values , since the centrifugal barrier in @xmath94-space is decided by the @xmath95-term in eq . .
for our cases of @xmath16 and @xmath17 these values are @xmath96 and @xmath97 , respectively .
detailed calculations of @xmath92 are described in methods . with these @xmath92
we obtain differential equations for @xmath98 from eq . , that is @xmath99 which are valid for strengths close to @xmath100 where @xmath101 . for @xmath16 ,
the large - distance behavior is a repulsive centrifugal barrier corresponding to two close - lying particles and the third far away in a state of relative angular momentum @xmath102 .
this barrier excludes infinitely many bound states . for @xmath15
the large - distance behavior is attractive with a leading term arising from two particles close to each other and the third far away with relative angular momentum @xmath31 . for @xmath103
an analytical solution can be found as in ref .
@xcite and easily proved by insertion into eq .
, that is @xmath104 where @xmath105 and the value of @xmath106 is related to the short - distance boundary condition .
this is an oscillating function of @xmath107 with an infinite number of nodes .
. has an infinite number of bound state solutions with @xmath108 .
each of these bound states of energy @xmath109 falls of exponentially when @xmath110 increases above @xmath111 .
however , for @xmath112 all solutions resemble @xmath113 , which therefore provides the estimate @xmath114 where @xmath115 obey @xmath116 .
thus , @xmath117 and we obtain a double exponential scaling . the simultaneous contributions from components , @xmath118 , are crucial for this conclusion .
the derivation of @xmath92 from solving the angular equation , eq .
, was very recently modified in ref . @xcite by including , in the angular wave function at small angles @xmath94 , the next order ( faster vanishing ) term in @xmath79 .
surprisingly , this leads to an additional slower vanishing , non - universal term , @xmath119 , which should be added in both radial equations eqs . and .
for details see methods .
this unusual behavior suggests that other higher order terms also should be investigated .
first , we emphasize that only the angular configurations with @xmath120 can contribute to a large - distance long - range behavior supporting infinitely many bound states .
second , non - adiabatic coupling terms in the radial equations are expected to vanish faster than the terms included so far . here
the diagonal coupling term @xcite , latexmath:[$q=\langle \phi_1|\frac{\partial^2}{\partial \rho^2 } significance for two reasons .
first , it is expected to be the slowest vanishing for large @xmath79 , and second omission or inclusion of this coupling in the lowest adiabatic potential produces , respectively a lower or upper bound for the exact binding energy @xcite .
the leading order result for @xmath122 is derived in methods to cancel exactly the non - universal term found in ref .
@xcite . as a byproduct @xmath123
is simultaneously demonstrated .
conclusions from these delicate considerations are first that the @xmath16 channel essentially is unaffected , since this very small additional term can not change the large - distance main contribution , @xmath124 , of the repulsive centrifugal barrier in eq . .
second , it can not change the short - distance attraction which is crucial for borromean structures .
third conclusion is that inclusion of these terms are very important for the @xmath17 channel , since it may change the large - distance behavior crucial for possible ( super ) efimov states .
the necessary rigorous mathematical derivations and accurate numerical investigations for @xmath17 are very involved and beyond the scope of this paper . however , if for instance , the next order of @xmath122 is equal to @xmath125 and the sum of all other couplings vanish faster , the prediction in ref .
@xcite would be confirmed .
we emphasize that this order in @xmath122 is crucial for the efimov effect as it provides both upper and lower bounds on the exact energy . at the two - body threshold as function of @xmath126 for a fixed value of @xmath127 corresponding to the first two particles fixed at positions @xmath128 indicated by the black dots .
the probabilities increases from light ( white ) to dark ( black ) colors .
the unit of length is @xmath33 . ]
the ambiguity in the @xmath17 channel is present but not as important for the @xmath16 channel , where the short - distance behavior is decisive . the effective potential in eq . obtained from eq .
is for @xmath129 simply the eigenvalue , @xmath130 of the kinetic energy operator @xmath131 , see ref .
the pauli principle selects @xmath132 for @xmath133 , respectively .
the potentials for small @xmath79 are then @xmath134 and @xmath135 . comparing the potentials at short and large distances we conclude that the effective potentials have to cross each other with more attractions in the @xmath16 channel at short distances .
this behavior is plotted schematically in fig .
[ fig - ad - pot ] which shows that the three - body ground state has @xmath16 .
let us now consider strengths , @xmath27 , such that @xmath136 , where @xmath137 is defined as the limit for binding three fermions .
since the wave function is fully antisymmetric we can use the result in ref .
@xcite to obtain a lower bound for the ground state energy , @xmath138 .
it then immediately follows that @xmath139 . as for three bosons in 3d
, the lower limit is reached for deep and narrow two - body potentials vanishing at large distance . to investigate three - body systems
numerically we apply the stochastic variational technique to the three - body schrdinger equation using the basis elements @xmath140 where @xmath141 is the operator that exchanges particle coordinates @xmath60 and @xmath142 , and @xmath143 are non - linear variational parameters that are found stochastically @xcite . in table
[ tab ] we present the three - body energies , @xmath144 , for the ground and doubly degenerate first excited states .
borromean binding is then allowed for strengths smaller than @xmath35 with corresponding smaller three - body binding energies .
away from @xmath35 the efimov condition of infinitely large scattering length is violated .
the ground state has @xmath16 for all the qualitatively different two - body potentials , and energies and sizes are of the same order as the potential depths and ranges .
the first excited states have much smaller binding energy , much larger extension , and quantum number @xmath15 .
they are the first in the possible sequence of super - efimov states .
the geometric structure of these states can be seen in the probability density , @xmath145 , shown in fig .
[ figground ] for potential @xmath49 .
we choose the @xmath146-coordinate to be @xmath147 with a particle distance equal to the root - mean - square value .
the structures in fig .
[ figground ] persist for the other potentials and for different values of @xmath148 .
the resulting triangular and linear chain structures resemble the ground and the celebrated first excited @xmath149 states for the @xmath94-cluster structure of the @xmath150c - nucleus @xcite . by decreasing @xmath27 from @xmath35 toward @xmath151
we move into the borromean window and away from infinite scattering length .
this quickly removes all possible @xmath15 excited bound states .
however , the @xmath16 ground state remains until the value @xmath152 which is calculated and given in tab .
energies and sizes for @xmath153 given as mean - square radii , @xmath154 , illustrate both stability and fragility of ground and excited states . ratios of their binding energies differ consistently with one order of magnitude , and excited state radii are much larger than for the ground state .
the effect of the outer barrier in @xmath53 is clearly seen through larger binding and smaller radii .
using a hyperspherical formalism in 2d we have found that three identical fermions support two very different types of bound states corresponding to angular momentum @xmath31 and @xmath102 .
we find a doubly degenerate excited state of angular momentum one for two - body potentials very close to the threshold for binding .
these states may be the lowest in an infinite sequence of super efimov states as predicted in ref .
our findings differ from ref .
@xcite where a non - universal term appears in the effective three - body potential .
however , we show that this term is cancelled precisely by inclusion of the leading order of the diagonal coupling . in any case , the possible states are fragile , extremely large and weakly bound , and observation of the related possible super efimov sequence would be correspondingly difficult .
a novel discovery is the existence of a borromean window defined as an interval in finite two - body strengths where no two - body subsystem is bound . within this window
the well bound zero angular momentum ground state is found to be located inside a barrier .
this is in sharp contrast to the system of three bosons @xcite .
numerical calculations for three spin - polarized systems are carried out for the first time in two dimensions using the coordinate space formalism with finite - range potentials .
the results confirm the analytically discussed properties for both ground and excited states .
these calculations can be used as a reference point for further numerical studies , that could be performed with atomic potentials obtained from state - of - the - art quantum chemical calculations .
the ground state borromean states are located behind an outer barrier as we have demonstrated here .
this implies that they should be observable as a peak in the atom loss rate when the state is located in the window for borromean binding precisely as the bosonic efimov effect has been observed @xcite . an alternative way to probe and populate
the states would be to use rf spectroscopy @xcite .
it is imperative to notice that we do not need the presence of feshbach resonances in the angular momentum one ( @xmath54-wave ) channel @xcite .
the method we have used to derive the results above assumes only that there is a short - range two - body potential between the fermions that can be tuned in some way so that one can approach the threshold for the existance of a two - body bound state between two identical fermions .
a @xmath54-wave feshbach resonance is one option , but it could also be achieved by using for instance a long - range interaction such as for instance a dipole - dipole force .
[ [ acknowledgements ] ] acknowledgements + + + + + + + + + + + + + + + + discussions with d .- w .
wang , j. levinsen , s. moroz , y. nishida and z. yu are gratefully acknowledged .
this work was funded by the danish council for independent research dff natural sciences and the dff sapere aude program .
here we discuss the behavior of the two - body binding energy near the threshold . at this point
we also would like to refer to some early works @xcite using jost function formalism . for our derivations
we use @xcite that the discrete spectrum of bound states with @xmath155 ( @xmath156 ) corresponds to the solutions of the equation @xmath157 where @xmath158 is the first hankel function of order one @xcite .
we focus on the ground state in the weak binding limit , @xmath159 , and write eq . in the form @xmath160 where @xmath161 is the euler - mascheroni constant and where we use the regular zero energy solution from eq .
( [ funct - small ] ) .
( [ jost - small ] ) with @xmath46 defines @xmath35 as the smallest solution of the equation : @xmath162 which from eq .
( [ aeq ] ) is seen to correspond to the solution of infinite scattering length , @xmath163 .
if @xmath27 is slightly larger than @xmath35 then the ground state is bound and the binding energy satisfies the equation @xmath164 where ( @xmath165 ) are positive constants .
this dependency was used to extract an accurate value of @xmath35 by fitting the numerically calculated @xmath12 as function of @xmath27 .
here we present the derivations that lead to the eqs.([eq - fin - rho1 ] ) and ( [ eq - fin - rho ] ) .
we show the approach to solve eq.([angular ] ) with the wave function in eq.([angfunc ] ) .
the set @xmath166 , where for simplicity we use @xmath167 and omit the related index , solves the following system of integro - differential equations @xcite @xmath168 where @xmath169 and the coupling term @xmath170 is defined as @xmath171 _ non - interacting case .
_ let us first consider the situation when @xmath39 , which also defines the extreme short - distance behavior of @xmath92 for potentials diverging slower than @xmath172 at zero .
in this situation the system of equations ( [ coupled_eq ] ) decouples @xmath173 where the superscript @xmath31 tells us that we work with free particles .
the solutions that are regular at @xmath174 are @xcite , @xmath175 where @xmath176 is the jacobi polynomial @xcite , @xmath177 is the normalization constant , and @xmath84 is a non - negative integer related to the eigenvalue @xmath92 by @xmath178 . this result can also be used to determine the behavior of @xmath92 near @xmath129 , since the right - hand side of eq.([coupled_eq ] ) contains the factor @xmath110 . from this
we learn that for @xmath17 the lowest centrifugal barrier is determined by @xmath179 ( since the antisymmetric wave function with @xmath180 and @xmath181 is zero ) . for @xmath16
the lowest barrier arises from @xmath182 . _
interacting case .
_ we aim to find the large - distance behavior of @xmath92 .
we introduce an angle @xmath183 , such that for @xmath184 eq.([coupled_eq ] ) is the non - interacting case ( right - hand side is zero ) with the regular boundary condition at @xmath185 and solved by @xmath186 where @xmath187 is given by @xmath188 .
this wave function diverges for @xmath189 , where the interacting solution has to be used . to obtain the solution we use that only the components with @xmath190 are necessary to lowest order , since higher partial waves are suppressed at small interparticle distances , i.e. @xmath191 , equivalent to large distances .
more precisely , in the absence of additional resonances in the higher partial wave channels with @xmath192 one can show that @xmath193 where @xmath84 is an integer ( corresponding to a free solution ) @xcite .
such terms will therefore vanish much faster than the @xmath190 terms and can thus be neglected .
_ case with @xmath17_. for @xmath17 we need to solve the following system of equations for small @xmath94 @xmath194 @xmath195 we note that @xmath196 , since they should yield the same @xmath197 , i.e. @xmath198 . for simplicity from now on we will write @xmath199
. now we expand the wave functions @xmath200 with @xmath201 in the vicinity of @xmath202 which leads to the coupling terms the homogeneous part of eqs.([system_eq ] ) and ( [ system_eq1 ] ) for small @xmath94 has the main contribution from the two - dimensional two - body equation , eq.([shr - rad - eq ] ) , as seen by using the substitution @xmath207 .
one solution to the inhomogeneous part is @xmath208 and @xmath209 . in this way
we find the solutions for @xmath191 @xmath210 where @xmath211 and @xmath212 are functions that up to normalization should be determined by matching the solutions and their first derivatives for @xmath184 and @xmath191 at @xmath213 .
this matching can be done only for specific values , @xmath197 , that can be be obtained from the following equation where we took the limit @xmath215 , since it defines the two - body threshold for binding .
to simplify this equation we use the following identities , see for example ref .
@xcite , @xmath216 @xmath217+\frac{\gamma(\nu+1)}{\pi\gamma(\nu+2)}\frac{1}{\alpha^2 } , \\ q_{\nu-1}^{(1,2)}&(\cos(2\alpha))\simeq\frac{\gamma(\nu+1)}{\pi\gamma(\nu ) } \bigg[-2\gamma+1-\psi_{\gamma}(\nu)- \nonumber \\ &
\psi_{\gamma}(\nu+2)-2\ln(\alpha)\bigg ] + \frac{\gamma(\nu+2)}{\pi\gamma(\nu+3)}\frac{1}{\alpha^2}\ ; , \end{aligned}\ ] ] where @xmath218 is the digamma function .
these identities allow eq.([eigen - eq ] ) to be rewritten @xmath219\right)+r_{11}\bigg ) \times \nonumber \\ \frac{\bigg(\frac{\gamma(\nu+1)}{\gamma(\nu)}\left(\cos(\pi\nu)+\frac{\sin(\pi\nu)}{\pi } [ \psi_{\gamma}(\nu)+2\ln\alpha_0]\right ) + r_{-1 - 1}\bigg)}{r_{-11}r_{1 - 1}}=-1 \nonumber \ ; , \label{eigen - eq1}\end{aligned}\ ] ] where we neglect terms smaller than @xmath220 in the limit of @xmath221 . the smallest solution to this equation is @xmath222 , with @xmath223 .
this solution yields @xmath224 , where @xmath225 . as we discuss in the main part of the paper this solution produces the infinite tower of states with the double exponential scaling .
_ validity of the derived large - distance behavior of @xmath197 at @xmath226 .
_ before applying this adiabatic potential we need to discuss its validity . to do so we first need to estimate what will be changed by adding higher order terms in eqs.([small_ang ] ) and ( [ small_ang1 ] ) ; and second we need to calculate the lowest order coupling term @xcite , @xmath227 where the averaging over all angles is taken for normalized angular wave functions , @xmath228
first , the higher order large - distance effects on the angular wave functions are necessary , since their neglect is the only assumption made after we decided to use only partial waves with @xmath190 .
second , inclusion of the diagonal coupling yields an upper bound for the exact binding energy , which together with the lower bound produced by @xmath197 establishes bounds for the exact three - body binding energy .
it was very recently shown @xcite that the higher order terms ( @xmath229 ) in the solution to the homogeneous angular part of eqs.([system_eq ] ) and ( [ system_eq1 ] ) contribute as @xmath230 to @xmath231 , where @xmath232 this surprising effect can be understood since the boundary conditions require that the constants in eqs.([small_ang ] ) and ( [ small_ang1 ] ) must satisfy @xmath233 .
this means that a higher order term in eqs.([system_eq ] ) and ( [ system_eq1 ] ) may be of the same order as the couplings neglected in eqs.([small_ang ] ) and ( [ small_ang1 ] ) . in ref .
@xcite it was pointed out that such an effect means that two - body observables alone is not sufficient to reproduce the correct asymptotic large - distance structure of @xmath197 .
thus , universal behavior is not guarantied . where @xmath236 is the two - body wave function decreasing as @xmath237 at infinity , the constants @xmath238 and @xmath239 are determined to satisfy normalization and boundary conditions at @xmath240 . after straightforward but tedious calculation we obtain @xmath241 .
we know @xcite that @xmath122 is negative , and consequently @xmath242 must be positive .
this can also easily be confirmed directly by rewriting @xmath242 as @xmath243 ^ 2\mathrm{d}x}{\lim_{x\to\infty}(xu(x))^2}.\ ] ] this means that the derived adiabatic potential necessarily should be supplemented with both the @xmath122 term and the term , @xmath119 , obtained by inclusion of next - to - leading order terms in the solution to the homogeneous part of eqs.([system_eq ] ) and ( [ system_eq1 ] ) .
it turns out that the leading contribution in the @xmath122 term cancels exactly the @xmath119 term .
thus , to make a rigorous conclusion about the allowed interval for the three - body binding energy , the next - to - leading order in @xmath122 term should be calculated .
this investigation is , however , out of scope of the present paper but it should be the focus in future more detailed and more elaborate work . _
case with @xmath16_. this case is not very different from the @xmath17 just discussed above . we therefore only give the resulting @xmath244 , and the couplings that are needed to solve the corresponding equations .
the couplings for @xmath245 take the form where @xmath248 defines @xmath244 as @xmath249 . in the same way as we obtained eq .
( [ eigen - eq ] ) we derive the following equation for @xmath248 @xmath250 this equation has a particular solution @xmath251 , which produces @xmath252 .
the corresponding adiabatic potential was presented in the main part of this paper .
we also note that to determine the large - distance behavior up to terms proportional to @xmath253 we need an investigation similar to the one provided for @xmath17 .
however , since the leading term in the adiabatic potential , @xmath124 , only marginally allows universal weakly - bound states the results of such calculations are not expected to be particularly interesting . in the @xmath16 channel
the most important physics comes from the non - universal adiabatic potential at short - distance .
v. efimov , yad .
fiz * 12 * , 1080 ( 1970 ) ; sov .
* 12 * , 589 ( 1971 ) . m. v. zhukov _
et al . _ ,
physics reports * 231 * , 151 ( 1993 ) .
richard and s. fleck , phys .
* 73 * , 1464 ( 1994 ) .
s. moszkowski _
et al . _ ,
a * 62 * , 032504 ( 2000 ) .
e. nielsen , d. v. fedorov , a. s. jensen , and e. garrido , physics reports * 347 * , 373 ( 2001 ) . a. s. jensen , k. riisager , d. v. fedorov , and e. garrido , rev .
phys . * 76 * , 215 ( 2004 ) .
e. braaten and h. w. hammer , phys
. rep . * 428 * , 259 ( 2006 ) .
n. a. baas _
et al . _ ,
physics of atomic nuclei , * 77 * , 361 ( 2014 ) .
e. nielsen , d. v. fedorov , and a. s. jensen , phys .
a * 56 * , 3287 ( 1997 ) .
a. g. volosniev , d. v. fedorov , a. s. jensen , and n. t. zinner , eur .
j. d * 67 * , 95 ( 2013 ) .
e. nielsen , d. v. fedorov , and a. s. jensen , few body syst .
* 27 * , 15 ( 1999 ) .
l. w. bruch and j. a. tjon , phys .
a * 19 * , 425 ( 1979 ) .
s. a. vugalter and g. m. zhislin , theor .
* 55 * , 493 ( 1983 ) .
a. g. volosniev : _ few - body systems in low - dimensional geometries _ , ( phd thesis , aarhus university , 2013 ) .
y. suzuki and k. varga : _ stochastic variational approach to quantum - mechanical few - body problems _
( springer , berlin , 1998 ) .
j. mitroy _
et al . _ , rev .
* 85 * , 693 ( 2013 ) .
t. kraemer _
_ , nature * 440 * , 315 ( 2006 )
. s. e. pollack , d. dries , and r. g. hulet , science * 326 * , 1683 ( 2009 ) .
m. zaccanti _
et al . _ , nature phys . * 5 * , 586 ( 2009 ) .
n. gross , z. shotan , s. kokkelmans , and l. khaykovich , phys .
lett . * 103 * , 163202 ( 2009 ) .
t. lompe , t. b. ottenstein , f. serwane , a. n. wenz , g. zrn , and s. jochim , science * 330 * , 940 ( 2010 ) .
s. nakajima , m. horikoshi , t. mukaiyama , p. naidon , and m. ueda , phys .
lett . * 106 * , 143201 ( 2011 ) .
o. machtey , z. shotan , n. gross , and l. khaykovich , phys .
lett . * 108 * , 210406 ( 2012 ) .
|
the study of quantum mechanical bound states is as old as quantum theory itself . yet
, it took many years to realize that three - body borromean systems that are bound when any two - body subsystem is unbound are abundant in nature . here
we demonstrate the existence of borromean systems of spin - polarized ( spinless ) identical fermions in two spatial dimensions .
the ground state with zero orbital ( planar ) angular momentum exists in a borromean window between critical two- and three - body strengths . the doubly degenerate first excited states of angular momentum one appears only very close to the two - body threshold .
they are the lowest in a possible sequence of so - called super - efimov states .
while the observation of the super - efimov scaling could be very difficult , the borromean ground state should be observable in cold atomic gases and could be the basis for producing a quantum gas of three - body states in two dimensions .
unlike classical mechanics quantum mechanics allows bound @xmath0-body states without having bound subsystems .
these so - called borromean systems are discussed in a number of publications for the simplest example of three particles @xcite .
the phenomenon was recently even generalized to more particles and higher orders , see f.ex .
ref .
@xcite .
borromean three - body systems are abundant in three dimensions ( 3d ) for both bosonic and fermionic systems in nuclear , atomic and molecular physics @xcite . however , the behavior is strikingly different for bosonic systems in one or two ( 2d ) spatial dimensions @xcite . where bound states appear for infinitesimally small attractions . without an artificial repulsive barrier at large distance
it is virtually impossible to form a borromean bosonic 2d system @xcite .
furthermore , the celebrated efimov effect of infinitely many bound three - body states at the two - body threshold is not present in 2d @xcite .
three identical spin - polarized fermions are harder to bind than bosons , because relative two - body @xmath1-states are forbidden by the pauli principle . it would therefore intuitively be more difficult to form borromean systems and the efimov effect should be out of reach .
however , both effects are possible in 2d as we shall discuss in the present paper .
the efimov effect was recently derived in an extreme double - exponential scaling form @xcite using a momentum space formalism .
the same scaling behavior was also found in low - energy scattering of three spinless fermions @xcite . in this paper
we present analytical and numerical evidence for a new state of fermionic matter where the basic constituent is a borromean three - fermion cluster .
this may be considered a trion quantum gas as opposed to the two - component bcs - bec crossover @xcite driven by the absence ( bcs ) or presence ( bec ) of a two - body bound state .
the new structure should be accessible through the tunability of both interactions and geometry of modern cold atom experiments @xcite .
both the single particles and the three - body borromean states are fermionic entities and thus have to obey the pauli principle
. this could imply an increase in stability of the many - body system , but estimates of the lifetime show that it is expected to be considerably shorter than the two - component fermi gas @xcite .
the creation of a superfluid state of spin - polarized fermions is still an important outstanding goal in the field .
observation of the three - body borromean ground state is , however , not dependent on reaching the superfluid state .
when formed , these three - body states are likely susceptible to chemical reactions and may be used to study quantum chemical dynamics of a fermionic gas with suppressed two - body but enhanced three - body reactions .
|
this work has been partially supported by the brazilian research agencies cnpq and fapesp .
baxter , `` exactly solved models in statistical mechanics '' . academic press , 1982 , new york c. fan and f.y .
. 179 ( 1969 ) 560 _ b.u .
felderhof , _ physica 66 ( 1973 ) 279 , 509 _ s. krinsky , _
phys.letters.a 39 ( 1972 ) 169 _ b. sutherland , _
j.math.phys 11 ( 1970 ) 3183 _ e. barouch , _ `` two - dimensional ferroelectric models '' , ed . c. domb and m.s . green , academic press , vol.1 , 1972 , page 366 _ p.w .
kasteleyn , `` exactly solved lattice models '' , in fundamental problems in statistical mechanics , vol.3 , egd cohen ed . , north holland , 1975 k. sogo , m. uchinami , y. akutsu and m. wadati , _ prog.theor.phys . 68 ( 1982 ) 508 _ v.v . bazhanov and yu.g .
stroganov , _ theor.math.phys . 62
( 1985 ) 253 _ s - k .
wang , _ j.phys.a:math.gen . 29 ( 1996 ) 2259 _ c. ahn , _
nucl.phys.b 422 ( 1994 ) 449 _ c. ahn and r.i .
nepomechie , _
nucl.phys.b 586 ( 2000 ) 611 _
|
in this note we study la baxter @xcite the possible integrable manifolds of the asymmetric eight - vertex model .
as expected they occur when the boltzmann weights are either symmetric or satisfy the free - fermion condition but our analysis clarify the reason both manifolds need to share a universal invariant .
we also show that the free - fermion condition implies three distinct classes of integrable models .
= -0.33 in = -0.3 in ufscar - th-02 + + _ departamento de fsica , universidade federal de so carlos _ _ caixa postal 676 , 13565 - 905 , so carlos , brazil _
july 2002 exactly solved vertex models play a fundamental role in classical statistical mechanics .
the most important of these is the so - called eight - vertex model which contains as special cases most systems on a plane square lattice @xcite .
the general asymmetric eight - vertex model possesses six different boltzmann weights @xmath0 and @xmath1 whose transfer matrix can be written as @xmath2\ ] ] where the trace is over the ordered product of local operators @xmath3 which are given by the following @xmath4 matrix @xmath5 and @xmath6 are pauli matrices acting on the sites @xmath7 of an one - dimensional lattice .
the asymmetric eight - vertex model is known to be solvable in the manifolds @xmath8 @xmath9 @xmath10 where @xmath11 and @xmath12 are arbitrary constants .
the manifold ( 3 ) is the so - called free - fermion model whose free - energy was first calculated by fan and wu @xcite and later re - derived by felderhof @xcite who devised a method to diagonalize the corresponding transfer matrix .
the integrability of the free - fermion manifold is usually assumed from the fact that its transfer matrix commutes with the @xmath13 hamiltonian as shown by krinsky @xcite who used a procedure first developed by sutherland @xcite .
later on barouch @xcite and kasteleyn @xcite have revisited the problem of commuting asymmetric eight - vertex transfer matrices and generalized heisenberg hamiltonians which leaded kasteleyn @xcite to point out the existence of the manifolds ( 4 ) . as stressed by this author , however , such manifolds are trivial because they can be seen as set of independent one - dimensional models and therefore they should be disregarded . on the other hand ,
the solution of the symmetric manifold was found by baxter through quite general approach denominated commuting transfer matrix method which culminated in the famous `` star - triangle '' relations @xcite .
the fact that the symmetric eight - vertex transfer matrix commutes with a related @xmath14 hamiltonian @xcite has then been made more precise because the latter is essentially a logarithmic derivative of the former @xcite .
it would be quite desirable to extend the baxter s method to the asymmetric eight - vertex model and to rederive the manifolds ( 3 ) and ( 5 ) from a unified point of view . since this approach does not assume a priori the existence of a specific local form for the corresponding hamiltonian it can lead us to new integrable manifolds not covered by the analysis of barouch @xcite and kasteleyn @xcite .
we recall that much of the work on this problem , see e.g. refs .
@xcite , has been concentrated to analyze the yang - baxter equations directly in terms of spectral parameters . though this is a valid approach it often hides the general integrable manifolds in terms of specific parameterizations which need to be found by a posteriori guess - work . a more direct way would be first to determine the solvable manifolds by an algebraic study la baxter of the corresponding `` star - triangle '' equations and afterwards to parameterize them by using the theory of uniformization of biquadratic polynomials @xcite .
it appears that kasteleyn @xcite was the first to make an effort toward such analysis but the best he could do was to guess the manifold ( 3 ) from known results by felderhof besides clarifying the origin of the pseudo one - dimensional manifold ( 4 ) as the linearization of the yang - baxter equation around a non - identity @xmath15 @xmath16-matrix . since the later possibility leads us to trivial manifolds we will disregard it , as did kasteleyn , from our forthcoming analysis .
the probable reason that such generalization has not yet been carried out seen technical since in the asymmetric model we have to deal with the double number of equations as compared to the symmetric eight - vertex model . at first sight
this appears to be a cumbersome task , but here we show that it is possible to simplify this problem , without recoursing to computer manipulations , to a number of simple equations that will clarify the common origin of the above two integrable manifolds . besides that , this approach allows to show that manifold ( 3 ) is one between three possible different integrable branches satisfying the free - fermion condition .
the `` star - triangle '' relations are sufficient conditions @xcite for commuting transfer matrices and for the asymmetric eight - vertex model they are given by @xmath17 @xmath18 @xmath19 @xmath20 @xmath21 @xmath22 note that each of these equations possesses two possibilities and we shall denote them by eqs.(@xmath23,@xmath24,@xmath25 ) .
altogether we have twelve linear homogeneous equations and only six weights , say @xmath26 and @xmath27 , are at our disposal to be eliminated in terms of remaining set of weights @xmath28 and @xmath29 .
therefore we have to choose the appropriate equations to start with and our solution goes as follows .
we first eliminate the weights @xmath30 with the help of the pair of equations ( @xmath31,@xmath32 ) and by substituting the result in eqs.(@xmath23 ) we find the following relations @xmath33\end{aligned}\ ] ] next we apply similar procedure in the case of eqs.(@xmath34,@xmath35 ) and the corresponding relations between the weights @xmath36 , @xmath37 and @xmath38 are @xmath39\end{aligned}\ ] ] from eqs.(@xmath40 ) it is not difficult to eliminate the weights @xmath36 , leading us to constraints between @xmath37 and @xmath38 @xmath41= d^{"}\left[a_{\mp}b_{\mp}f(a_{\pm}^{'},b_{\pm}^{'},c^{'},d^{'})-a_{\pm}^ { ' } b_{\pm}^{'}f(a_{\pm},b_{\pm},c , d ) \right]\ ] ] at this point it is tempting to use such equations and the previous results for @xmath30 and @xmath36 to eliminate five weights ratios and to substitute them in the remaining equations , namely eqs.(@xmath25 ) and either eqs.(@xmath42 ) or eqs.(@xmath43 ) .
this is , however , not so illuminating because it leads us to carry out simplifications in complicated expressions .
we find that it is more profitable to repeat the procedure described above but now we first eliminate the weights @xmath44 and in the end we use eqs.(@xmath25 ) instead of eqs.(@xmath23 ) .
this leads us to a different constraint between @xmath37 and @xmath38 given by , @xmath45= c^{"}\left[a_{\mp}b_{\mp}f(a_{\pm}^{'},b_{\pm}^{'},c^{'},d^{'})-a_{\mp}^ { ' } b_{\mp}^{'}f(a_{\pm},b_{\pm},c , d ) \right]\ ] ] now we reached a point that enables us to make conclusions on the way the set of weights @xmath46 and @xmath47 should be related to each other .
in fact , from eqs.(@xmath48,@xmath49 ) we find that the necessary conditions for the weights @xmath37 and @xmath38 not to be all zero are @xmath50 and either @xmath51 or @xmath52 we are already in the position to conclude that the asymmetric eight - vertex model has indeed only two possible integrable manifolds , one is singled out by the free - fermion condition ( 17 ) while the other ( 18 ) turns out to be a mixed type of conditions that relate the set of weights both alone and between each other .
one important point of our analysis is that it makes clear that both manifolds need to share a common invariant given by eq.(16 ) to close our analysis it remains to check the consistency between eqs.(@xmath23 ) and eqs.(@xmath25 ) which can in principle be a source of further constraints . from such equations
one can easily calculate the ratios @xmath53 and @xmath54 , namely @xmath55 @xmath56 which in principle can be compared with our previous results for the same ratios . before proceeding with that ,
however , there exists one property that we have not yet explored . instead of starting our analysis by eliminating the weights @xmath57,@xmath36 , @xmath37 and @xmath38 we could choose to begin with the other two sets of weights as well .
because the `` star - triangle '' equations are not symmetric by exchanging a given two sets of weights we expect that each possibility will leads us to different kind of constraints .
this means that we can use the asymmetry of the weights in our favour which may help us in further simplifications .
for example , the relations ( @xmath23-@xmath25 ) are invariant under the exchange of weights @xmath58 and @xmath59 only after the transformation @xmath60 is performed for all set of weights .
this means that if we had started our procedure by eliminating the weights @xmath61 and @xmath1 the same analysis we have carried out so far will lead us to the following constraints @xmath62 besides that either @xmath63 or @xmath64 by the same token if we had started by eliminating @xmath65 and @xmath66 we will find @xmath67 and that either @xmath68 or @xmath69 let us now analyze the consequences of this observation for each possible integrable manifold and here we begin with the second manifold .
it is not difficult to see that the consistency of the equations ( 18),(23 ) and ( 26 ) , to what concern relations within the same set of weights , impose severe restrictions on the second type of the manifold , namely @xmath70 and similar conditions for the other sets @xmath71 and @xmath72 .
it turns out , however , that the only possibility compatible with the `` universal '' constraints ( 16 ) , ( 21 ) and ( 24 ) is the totally symmetric case @xmath73 and @xmath74 leading us therefore to the baxter s model ( 5 ) .
note that in this situation the compatibility between eq(@xmath23 ) and eq.(@xmath25 ) is trivial because both equations ( 19 ) and ( 20 ) are automatically satisfied .
we now turn our attention to the free - fermion manifold . in this case
we have much less restrictive constraints since we are only left with relations between different weights , namely eqs.(16,21,24 ) .
altogether these equations provide us the following relation @xmath75 whose compatibility with eqs.(@xmath23,@xmath25 ) can be implemented by evaluating the left - hand side of eq.(28 ) with the help of eqs.(19,20 ) .
after few manipulations , in which the free - fermion condition is explicitly used , we end up with a `` separable '' equation @xmath76 for the weights @xmath59 and @xmath77 and the polynomial @xmath78 is given by @xmath79 \nonumber\\ & & \times \left [ a_{-}b_{+}a_{-}^{'}b_{+}^{'}-a_{+}b_{-}a_{+}^{'}b_{-}^ { ' } \right ] \nonumber\\ & & \times \left [ ( a_{+}^{2}+b_{-}^2-a_{-}^2-b_{+}^2)(a_{+}^{'}b_{-}^ { ' } + a_{-}^{'}b_{+}^ { ' } ) - ( a_{+}b_{- } + a_{-}b_{+ } ) ( { a_{+}^{'}}^{2}+{b_{-}^{'}}^2-{a_{-}^{'}}^2-{b_{+}^{'}}^2 ) \right ] \nonumber\\\end{aligned}\ ] ] from this equation we conclude that we have three possible free - fermion integrable manifolds given by either @xmath80 or @xmath81 or still @xmath82 besides of course the free - fermion conditions for both @xmath59 and @xmath83 together with the `` universal '' relation ( 16 ) .
note that the free - fermion case ( 31 ) can not be related to the manifold ( 4 ) beginning by the fact that in the former model the weight @xmath84 can be different of the weight @xmath1 . in figure 1
we have summarized all the results obtained so far .
let us now compare our results with previous work in the literature .
contrary to what happened to the symmetric manifold ( 5 ) we recall that eqs.(30,31 ) do not imply that the ratios @xmath85 and @xmath86 are necessarily constants but only that they are invariants for two distinct set of weights .. ] this is the reason why general solutions of the yang - baxter equation satisfying the free - fermion condition are expected to be non - additive @xcite .
in fact , in the appendix @xmath87 we show that the additional assumption of additivity provides us extra restriction to the weights . in this sense ,
the manifold ( 30 ) turns out to be a generalization of the original result ( 3 ) by krinsky @xcite .
next the manifold ( 31 ) has been only partially obtained in the literature , more precisely in the special case @xmath73 and @xmath88 @xcite and @xmath74 is contained in the baxter s solution . ] .
finally , to the best of our knowledge the last branch ( 32 ) is new in the literature .
the probable reason that such general manifolds have been missed in previous work , see for example refs.@xcite , is related to analysis of the yang - baxter equation in terms of spectral parameters .
there it was required that at certain value of the spectral parameter ( initial condition ) the weights should be regular , i.e that the corresponding @xmath89 operator be proportional to the four dimensional permutator .
note that the @xmath89 operator of manifold ( 32 ) can not be made regular and therefore does not have a local associated hamiltonian .
this is also the reason barouch @xcite and kasteleyn @xcite missed such manifold since they used the assumption of local forms of hamiltonians .
we recall that though the property of regularity guarantees that the logarithmic derivative of the transfer matrix is @xmath90 this is by no means a necessary condition for integrability . in summary
, we have analyzed according to baxter the integrable branches of the asymmetric eight - vertex model . besides recovering the baxter s model we shown that the free - fermion condition produces three different set of integrable manifolds . a natural question to be asked
is whether or not the new manifolds ( 31 ) and ( 32 ) can be solved by the method devised by felderhof originally proposed to diagonalize the transfer matrix of the krinsky s manifold ( 3 ) .
this is of interest since these systems maybe the corner stone of highly non trivial models as recently have been discussed in refs.@xcite .
in fact , we have evidences that the manifold ( 31 ) is related to a staggered @xmath13 model . because both the baxter symmetric model and the free - fermion manifolds ( 30 - 32 ) share a common algebraic structure , the yang - baxter algebra , it is plausible to think that baxter s generalized bethe ansatz can be adapted to include the solution of the free - fermion models too .
this problem has eluded us so far though some progress has been made in the case of the simplest free - fermion branch ( 31 ) .
* appendix a * the purpose here is to demonstrate that the hypothesis of additivity of the weights leads us to much more restrictive conditions for the free - fermion manifold as compared with the results ( 30 - 32 ) of the main text . in order to see that lets us consider as usual that the weights @xmath91 are parameterized by the variables @xmath92 and @xmath93 and similarly that @xmath94 and @xmath95 are parameterized by @xmath96 and @xmath97 , respectively .
the consistency between the `` universal '' relations ( 16 ) , ( 21 ) and ( 24 ) implies a remarkable separability condition for the ratio @xmath98 where @xmath99 is an arbitrary function . the additional assumption that the weights are additive means that this function is necessarily a constant which ultimately leads us to the relation @xmath100 as a consequence of that the possible manifolds satisfying the the free - fermion condition are either @xmath73 or @xmath74 .
now by imposing the consistency between eqs(@xmath23 ) and eqs.(@xmath25 ) it turns out that these two possibilities becomes either @xmath101 or @xmath102 where @xmath103 is a constant
. clearly , these are special cases of the manifolds ( 31 ) and ( 30 ) , respectively .
|
the hamiltonian road to quantization has played a key role , over the last century , in the development of quantum mechanics @xcite , quantum field theory in flat spacetime @xcite , including quantum yang - mills @xcite and the particle physics standard model , as well as in the formulation of canonical quantum gravity @xcite .
the main drawback of the hamiltonian formulation , despite its beautiful and powerful applications to the classical cauchy problem of general relativity @xcite , lies in the loss of the full diffeomorphism group of four - dimensional spacetime , with the associated undoing of the unification of space and time into the spacetime manifold ( see , however , the valuable work in refs .
@xcite on the way to circumvent this problem ) . indeed , at classical level , the tools of global differential geometry make it possible to obtain a spacetime covariant formulation of the constraint equations @xcite , which turn out to be linear @xcite on a bigger space , the space of multimomenta .
even earlier , at quantum level , the work of peierls @xcite and dewitt @xcite made it possible to define a poisson bracket on the space of all field functionals that remain invariant under the action of the infinite - dimensional lie ( pseudo-)group of the theory ( for gravity , this is the group of spacetime diffeomorphisms ) .
if @xmath0 and @xmath1 are any two such functionals of the field variables @xmath2 , their classical peierls bracket reads as ( see appendix for the notation ) @xmath3 where @xmath4 , the supercommutator function @xcite , is the difference between advanced and retarded green functions for the invertible operator @xmath5 acting on fields : @xmath6 the advanced and retarded green functions are both left- and right - inverses of @xmath5 , i.e. @xmath7 @xmath8 in the framework of ref .
@xcite , the state of the art on the application of such ideas to einstein s gravity was as follows . 0.3 cm ( i ) the absolute invariants ( also called observables or gauge - invariant functionals ) @xmath0 and @xmath1 satisfy , by definition , the conditions @xmath9 which are a particular case of the gauge - invariance equation @xmath10 of appendix a ( see ( a5 ) ) , with @xmath11 , because the infinitesimal diffeomorphism @xmath12 is a particular case of the general equation @xmath13 of appendix a ( see ( a3 ) ) . 0.3 cm ( ii ) the infinitesimal variation suffered from @xmath0 , say , reads as @xmath14 since the spacetime metric is not an invariant , its variations @xmath15 are determined only up to a coordinate transformation ( see ( 1.6 ) ) , while eqs . (
1.5 ) guarantee that the advanced and retarded variations @xmath16 do not suffer from redundancies .
0.3 cm ( iii ) if @xmath17 is the einstein - hilbert action , the associated euler - lagrange equations are ( hereafter @xmath18 ) @xmath19 if the action functional undergoes the change @xmath20 with constant parameter @xmath21 , the new euler - lagrange equations read as @xmath22 the general solution of eq .
( 1.10 ) is obtained by adding ( 1.6 ) to a particular solution determined by appropriate boundary and supplementary conditions .
a convenient form of supplementary ( or gauge - fixing ) condition is @xcite @xmath23 if eq . ( 1.11 ) is fulfilled , the euler - lagrange equations ( 1.10 ) take the form @xmath24\delta^{\pm}g_{\sigma \tau } = -2 \varepsilon { \delta b \over \delta g_{\mu \nu}}. \label{(1.12)}\ ] ] one can then solve for the infinitesimal variation of the spacetime metric in the form @xmath25 where @xmath26 are the advanced and retarded green functions for eq .
( 1.12 ) ; they satisfy @xmath27 with the notation of ref .
@xcite , the indices @xmath28 refer to the point @xmath29 while the indices @xmath30 refer to the point @xmath31 .
parallel displacement along the geodesic between @xmath31 and @xmath29 is performed by the bivector @xmath32 .
the four - dimensional dirac delta is a scalar density at both @xmath31 and @xmath29 . by construction ,
the green functions @xmath26 are bi - tensors ( see the beginning of sec .
iii ) , by virtue of their transformation properties at two different spacetime points . 0.3 cm ( iv ) in canonical quantum gravity , inspired by the dirac map in quantum mechanics @xcite , according to which the commutator of position and momentum operators is the imaginary unit times their classical poisson bracket , dewitt considered the commutator defined by @xmath33 \equiv { \rm i } \int { \rm d}^{4}x \int { \rm d}^{4}z { \delta a \over \delta g_{\mu \nu}}(x ) { \widetilde { \bf g}}_{\mu \nu \alpha \beta}(x , z ) { \delta b \over \delta g_{\alpha \beta}}(z ) , \label{(1.15)}\ ] ] where @xmath34 is the supercommutator for quantum einstein s gravity , i.e. @xmath35 and boldface characters are here used to stress that we deal with the operator counterpart of classical green functions .
the work in ref .
@xcite pointed out that , if it were possible to ignore the noncommutativity of factors on the right - hand side of eq .
( 1.15 ) , this ( formal ) commutator can be shown to satisfy all identities of a quantum poisson bracket , and hence the question of a consistent quantum theory of gravity was reduced ( but not solved ! ) to finding an appropriate definition of operator propagator .
we are here initiating a research program aimed at showing that one can get pretty close to fulfilling such a task after a careful consideration of well - known properties of classical hyperbolic equations .
for this purpose , sec .
ii introduces fundamental solution and parametrix of a scalar wave operator ; tensor wave equations are considered in sec .
iii , with emphasis on maxwell theory in curved spacetime ; nonlinear wave equations are studied in sec .
iv , by focusing on the choquet - bruhat method for solving cauchy s problem for the einstein equations .
v proves the existence of the parametrix necessary to solve the classical einstein equations , by using a well - known technique for proving the existence of a parametrix for elliptic ( rather than hyperbolic ! ) equations .
section vi studies the role of the parametrix in the jacobi identity for the peierls bracket , sec .
vii obtains novel functional equations for the parametrix , while sec .
viii discusses the remainder in the supercommutator function and presents our concluding remarks .
relevant background material is described in the appendices : dewitt s notation for gauge theories , spacetime geometry for the wave equation , characteristic conoid and kirchhoff formulas for the linear hyperbolic system associated to the nonlinear einstein equations .
let us consider , for simplicity , a scalar differential operator @xmath36 with @xmath37 coefficients @xmath38 that is defined on a connected open set @xmath39 of four - dimensional spacetime .
a _ fundamental solution _ of @xmath36 is a distribution @xmath40 which is a function of the spacetime point @xmath41 such that @xmath42 this means that @xcite @xmath43 we note that the fundamental solution is not uniquely defined , unlike the case of green s functions @xcite , for each of which there is a specific boundary condition .
a @xmath37 _ parametrix _ of @xmath36 is instead a distribution @xmath44 such that @xcite @xmath45 this concept is of interest because , although a fundamental solution is a useful tool for solving linear partial differential equations , some of the problems in which it plays a role can be handled better by means of functions possessing a singularity that is not annihilated but merely smoothed out by the differential operator under investigation .
the smoothing might even be so weak that the singularity is actually augmented but acquires a less rapid growth than was to be anticipated from the order of the differential operator @xcite . if one studies the scalar wave equation on minkowski spacetime in the coordinates @xmath46 , with @xmath47 , the solution of the cauchy problem @xmath48 can be expressed in the form @xmath49 where , on denoting by @xmath50 the fourier transforms of cauchy data , the operators @xmath51 act @xcite in such a way that ( here @xmath52 ) @xmath53 where the @xmath54 are real - valued _ phase functions _ which satisfy the initial condition @xmath55 and @xmath56 is a regularizing operator ( see our earlier comments ) which smoothes out the singularities acted upon by it @xcite .
thus , the amplitudes @xmath57 and phases @xmath54 make it possible to build the parametrix for the cauchy problem . in our analysis
we are going to need the generalization of this construction to the tensor wave equation , bearing in mind that eq .
( 1.14 ) is the tensor version of eq .
. meanwhile , let us try to develop a heuristic argument on the relation between a fundamental solution @xmath58 and a @xmath37 parametrix @xmath59 of a partial differential operator @xmath36 as in ( 2.1 ) . by omitting , for simplicity of notation , the subscript @xmath41 ,
if we consider the split @xmath60 we find @xmath61 thus , a fundamental solution of @xmath36 differs from its @xmath37 parametrix by minus the inverse of @xmath36 applied to the regularizing term @xmath62 . in the simplest case , i.e. operators on the real line , we would therefore solve the inhomogeneous equation @xmath63 in the form @xmath64 now , to the extent that the parametrix leads to a good approximate inverse of @xmath36 , the integral @xmath65 on the right - hand side of ( 2.11 ) provides a good approximation of @xmath66 similarly , in the tensor wave equations which are the ultimate goal of our investigation , with the associated peierls bracket , we are going to re - express ( 1.1 ) as the sum of the parametrix and remainder contribution , respectively .
the difference between fundamental solution and parametrix is further elucidated , with the notation summarized in appendix b , by sections 4.2 and 4.3 of ref .
@xcite , which prove that there exist fundamental solutions of the operator @xmath36 having the form ( see ( b1)-(b6 ) ) @xmath67 as well as parametrices reading as @xmath68 where @xmath69 is a smooth function belonging to @xmath70 , expressible , for which @xmath71 , while the world function @xmath72 takes the form @xmath73 . ] by @xmath74 the @xmath75 being the covariant components of the vector @xmath76 in ( 2.1 ) , while @xmath77 solves the characteristic initial value problem @xmath78 and is real - analytic , hence expressible as @xmath79 the @xmath80 being elements of @xmath70 computable in the form @xmath81 while @xmath82 where @xmath83 and takes values @xmath84 the @xmath85 being a sequence of positive numbers , strictly increasing and tending to infinity . by virtue of ( 2.12)-(2.18 ) , one finds ( cf .
( 2.10 ) ) @xmath86 where @xmath87 note that the parametrix is not determined uniquely . if @xmath88 and @xmath89 when the world function @xmath90 , then @xmath91 on @xmath92 , the null cone of @xmath41 , and @xmath93 when @xmath90 .
hence one can replace @xmath94 with @xmath95 in the formulas ( 2.13 ) , to obtain another pair of parametrices of @xmath36 in @xmath39 .
if @xmath39 is a causal domain ( see appendix b ) , the operator @xmath36 in ( 2.1 ) has a fundamental solution @xmath96 in @xmath39 , such that @xmath97 it is of the form @xcite @xmath98 with @xmath99 defined as in ( 2.14 ) , whereas , given the set @xmath100 the function @xmath101 is of class @xmath37 on @xmath102 and has support contained in @xmath102 .
when the world function @xmath103 for @xmath104 , the hadamard series on the right - hand side of ( 2.16 ) is an asymptotic expansion for @xmath105 . in analogous way , @xmath36 has also a fundamental solution @xmath106 in @xmath39 such that @xmath107 which is of the form @xmath108 where @xmath109 has properties similar to those of @xmath105 , with past and future interchanged .
we know that there are infinitely many @xmath37 parametrices of @xmath36 supported in the causal future of @xmath41 , and the previous theorem shows that one can construct a fundamental solution @xmath96 from any one of these .
however , another theorem ensures that there is in fact _ only one forward fundamental solution _
@xcite , which is precisely the advanced green function needed in the construction of the peierls bracket .
thus , eq . ( 2.23 ) is the definitive form of the forward fundamental solution of @xmath36 in @xmath39 .
it may be viewed as the field of a point source at @xmath41 .
it consists of a singular part @xmath110 which is a measure , supported on the future null cone of @xmath41 , and of a regular part @xmath105 , which is a function .
the function @xmath111 has its support contained in @xmath112 , and lies in @xmath113 .
it is said to be the _ tail _
term of the fundamental solution ; it distinguishes the curved spacetime case from the minkowskian case , where @xmath96 is sharp , its support being the future null cone of @xmath41 .
the tail term is a solution of the characteristic initial - value problem @xmath114
the tensor analogue of the differential operator @xmath36 in ( 2.1 ) can be written , by introducing tensor multi - indices @xmath115 and @xmath116 , in either of the forms @xcite @xmath117 or @xmath118 the class of @xmath37 tensor fields of rank @xmath119 defined on a causal domain @xmath120 is a vector space over the complex numbers , denoted by @xmath121 .
the subspace of @xmath121 consisting of fields with compact support is denoted by @xmath122 .
a _ tensor distribution _
@xmath123 is then a continuous linear map @xmath124 .
moreover , if @xmath125 is a point of @xmath120 , a _
tensor - valued distribution _
@xmath126 is a continuous linear map @xmath127 of @xmath122 into the finite - dimensional vector space of tensors of rank @xmath119 at @xmath125
. a _ bitensor _ of type @xmath128 at @xmath129 and of type @xmath130 at @xmath125 is a multilinear form on the product of @xmath119 copies of the cotangent space @xmath131 and of @xmath132 copies of the tangent space @xmath133 .
bitensor indices can be raised and lowered by means of the appropriate components of the metric tensor at @xmath129 and at @xmath125 , respectively .
the dirac distribution of rank @xmath119 , denoted by @xmath134 , maps @xmath135 to @xmath136 , according to @xmath137 a fundamental solution of the differential operator ( 3.1 ) or ( 3.2 ) is a tensor - valued distribution @xmath138 such that @xmath139 there exist two basic fundamental solutions @xmath140 and @xmath141 whose supports are contained in @xmath142 and in @xmath143 , respectively .
the _ transport bitensor _
( see @xmath32 in ( 1.14 ) ) in a geodesically convex domain @xmath39 is a bitensor field @xmath144 of rank @xmath119 at both @xmath125 and @xmath129 which satisfies the differential equations ( hereafter the world function @xmath145 , and the covariant derivatives @xmath146 and @xmath147 act at @xmath129 ) @xmath148 jointly with the initial conditions @xmath149 we shall also need the biscalar @xmath150 , which solves the differential equation @xcite @xmath151 with initial condition @xmath152 and is given , in local coordinates , by @xcite @xmath153 both @xmath154 and @xmath155 are symmetric functions of their arguments .
we can now state the tensor analogue of the theorem at the end of sec .
ii for the forward fundamental solution @xcite .
0.3 cm * theorem * in a causal domain @xmath120 , the tensor differential operator @xmath156 has a fundamental solution @xmath140 in @xmath120 such that @xmath157 which is of the form @xmath158 where @xmath159 is a bitensor field that vanishes if @xmath160 and is of class @xmath37 on the closed set @xmath161 now we remark that vacuum maxwell theory in the absence of charges and currents is ruled precisely by a tensor differential operator of the kind ( 3.10 ) , because , upon imposing the lorenz @xcite supplementary ( or gauge - fixing ) condition , i.e. @xmath162 the potential @xmath163 obeys the wave equation @xmath164 where @xmath165 having denoted by @xmath166 the dalembert operator @xmath167 , and by @xmath168 the ricci tensor of the spacetime manifold @xmath169 endowed with a levi - civita connection @xmath170 .
so far , we have met the concepts of fundamental solution and parametrix for linear partial differential equations , of either scalar or tensorial nature . but how to define fundamental solutions for the nonlinear partial differential equations provided by einstein s theory of gravitation ? indeed , the vacuum einstein equations in four spacetime dimensions are equivalent to ricci - flatness , i.e. @xmath171 , where , upon defining @xmath172 one has the split @xmath173 having defined @xcite @xmath174 @xmath175 where @xmath176 is a polynomial of covariant and contravariant metric components , and of their first partial derivatives . equations ( 4.1)-(4.3 ) suggest considering the supplementary condition @xmath177 , so that the vacuum einstein equations read eventually as @xmath178 now if we set @xmath179 we realize that our gauge - fixed vacuum einstein equations take the standard form of quasilinear hyperbolic systems @xcite @xmath180 suppose now , following again ref .
@xcite , that in a spacetime domain @xmath181 , centred at the point @xmath182 with coordinates @xmath183 , and defined by @xmath184 and for values of the unknown functions @xmath185 and their first derivatives satisfying @xmath186 the coefficients @xmath187 and @xmath188 possess partial derivatives with respect to their arguments up to the fifth order .
one can now show that , by differentiating @xmath189 times the eqs .
( 4.7 ) with respect to the variables @xmath190 , one obtains a linear system of second - order partial differential equations @xmath191 where the associated quadratic form @xmath192 is of normal hyperbolic type , i.e. @xmath193 in the following calculations , it is useful to define @xmath194 until we denote by @xmath195 the partial derivatives of fifth order of @xmath185 , i.e. @xmath196 differentiation of eqs .
( 4.7 ) with respect to any variable @xmath190 whatsoever leads to @xmath197 equations having the form @xmath198{\partial w_{s \mu}\over \partial x^{\lambda } } \nonumber \\ & + & { \partial f_{s}\over \partial x^{\alpha } } + { \partial f_{s}\over \partial w_{r}}w_{r \alpha } + { \partial f_{s}\over \partial w_{r \nu}}{\partial w_{r \nu}\over \partial x^{\alpha}}=0 . \label{(4.14)}\end{aligned}\ ] ] by repeating four times this process , the following system of @xmath199 equations is obtained : @xmath200 { \partial \over \partial x^{\lambda}}w_{s \beta \gamma \delta \varepsilon \mu } \nonumber \\ & + & \left[{\partial a^{\lambda \mu}\over \partial x^{\beta } } + { \partial a^{\lambda \mu}\over \partial w_{r}}w_{r \beta } + { \partial a^{\lambda \mu}\over \partial w_{r \nu}}w_{r \nu \beta}\right ] { \partial \over \partial x^{\lambda}}w_{s \alpha \gamma \delta \varepsilon \mu } + ... \nonumber \\ & + & \left[{\partial a^{\lambda \mu}\over \partial x^{\varepsilon } } + { \partial a^{\lambda \mu}\over \partial w_{r}}w_{r \varepsilon } + { \partial a^{\lambda \mu}\over \partial w_{r \nu}}w_{r \nu \varepsilon}\right ] { \partial \over \partial x^{\lambda}}w_{s \alpha \beta \gamma \delta \mu } \nonumber \\ & + & { \partial a^{\lambda \mu}\over \partial w_{r \nu } } { \partial w_{r \nu \alpha \beta \gamma \delta}\over \partial x^{\varepsilon } } + { \partial f_{s}\over \partial w_{r \nu } } { \partial w_{r \nu \alpha \beta \gamma \delta}\over \partial x^{\varepsilon } } + f_{s}=0 , \label{(4.15)}\end{aligned}\ ] ] where @xmath201 is a function of the variables @xmath190 , the unknown functions @xmath185 and their partial derivatives up to the fifth order included , but not of their higher - order derivatives . by virtue of eq .
( 4.15 ) , the fifth derivatives @xmath195 of the functions @xmath185 satisfy , within the domain @xmath181 and under the assumptions previously stated , a system of @xmath199 equations @xmath202 which is precisely of the linear type studied in ( 4.10 ) , because @xmath203 .
\label{(4.17)}\end{aligned}\ ] ] by virtue of the method described in appendix c , we know that eqs .
( 4.16 ) can be solved explicitly through kirchhoff - type formulas as in eq .
thus , by virtue of ( 4.13 ) , if we can find a fundamental solution , or at least a parametrix @xmath59 of the fifth - order linear equation @xmath204 @xmath195 being given by the integral formula ( c33 ) , we will know , at least in principle , the classical peierls bracket for gravity , which is in turn necessary to arrive at the desired definition of quantum peierls bracket ( 1.15 ) for gravity .
we recall that the @xmath185 functions in eq .
( 4.18 ) are the covariant components of the metric tensor of the spacetime manifold .
they can be expressed , from eq .
( 4.18 ) and its parametrix , through the convolution formula @xmath205
we can point out that , for every choice @xmath206 of the indices @xmath207 , we can set @xmath208 and hence the operator on the left - hand side of eq .
( 4.18 ) may be viewed as a constant - coefficient fifth - order elliptic operator on @xmath209 . at this stage
, we can exploit the following theorem @xcite : 0.3 cm * theorem*. every elliptic operator @xmath210 with constant coefficients has a parametrix which is a @xmath37 function in @xmath211 .
0.3 cm * proof*. the symbol of @xmath210 can be written as @xmath212 where the term @xmath213 is homogeneous of degree @xmath214 , and @xmath215 when @xmath216 .
then there exists a constant @xmath217 for which @xmath218 when @xmath219 , so that the homogeneity provides the minorization @xmath220 by virtue of ( 5.3 ) , one can write for some constants @xmath221 and @xmath222 @xmath223 provided that @xmath224 has @xmath225 .
since a derivative of order @xmath226 of @xmath227 is of the form @xmath228 , with @xmath229 of degree @xmath230 , as one can prove by induction , one finds that , when @xmath231 , @xmath232 choose now @xmath233 equal to @xmath234 in the subset of @xmath235 given by @xmath236 .
this ensures that @xmath237 is a bounded @xmath37 function , hence it can be viewed as the fourier transform of a distribution @xmath59 .
hence one has @xmath238 where the fourier transform of @xmath239 is @xmath240 .
hence @xmath239 itself is an element of the schwartz space , and it follows from ( 5.5 ) that @xmath241 is continuous when @xmath242 .
0.3 cm as is stated in ref .
@xcite , the error term @xmath239 obtained in such a proof has an analytic extension to @xmath243 .
moreover , the parametrix @xmath59 admits an analytic extension to a conic neighborhood of @xmath211 .
inspired by the split ( 2.9 ) , let us first re - express the classical advanced and retarded green functions mentioned in the introduction in the form @xmath244 @xmath245 where @xmath246 is the parametrix and @xmath247 is the remainder , as in ( 1.3 ) and ( 1.4 ) , to stress that the indices @xmath248 refer to different spacetime points , but we refrain ourselves from doing so for simplicity of notation . ] term .
moreover , inspired by the definition ( 1.2 ) , let us introduce @xmath249 the lack of uniqueness of the parametrix ( see comments after ( 2.21 ) ) can be exploited to ensure that @xmath250 has the same antisymmetry property of the supercommutator function @xmath4 . in other words ,
given the parametrices @xmath251 , suppose that @xmath252 } , \label{(6.4)}\ ] ] where @xmath253 and @xmath254 } \not = 0 $ ] .
now we can add to @xmath255 a function @xmath256 , tending to @xmath257 when @xmath90 , such that @xmath258 and hence it is enough to choose @xmath259 to achieve the desired skew - symmetry of @xmath250 .
the remainder term @xmath260 is then skew - symmetric as well , by virtue of ( 6.3 ) . in the proof that the classical peierls bracket is a poisson bracket on the space of gauge - invariant functionals , a crucial role is played by the verification of the jacobi identity @xcite obtained from the sum @xmath261 , \label{(6.7)}\end{aligned}\ ] ] where @xmath262 is found to vanish classically , as shown in detail in ref .
@xcite and , first , in ref . @xcite . by virtue of the split ( 6.3 ) ,
the jacobi identity ( 6.7 ) can be written as the sum of three terms , i.e. @xmath263 where @xmath264 ( respectively , @xmath265 ) can be obtained from ( 6.7 ) upon replacing everywhere @xmath4 with @xmath250 ( respectively , @xmath260 ) , whereas @xmath266 denotes the sum of all mixed terms , e.g. @xmath267 let us here focus on the term @xmath268 in the classical jacobi identity , because , within our scheme , it is equal to the classical part of the quantum jacobi identity . by virtue of ( 6.3 ) and
( 6.7 ) , we find @xmath269 the skew - symmetry of @xmath250 , jointly with commutation of functional derivatives : @xmath270 for all @xmath271 , implies that the first three terms on the right - hand side of ( 6.9 ) vanish .
for example , one finds @xcite @xmath272 and an identical procedure can be applied to the terms containing the second functional derivatives @xmath273 and @xmath274 in ( 6.9 ) .
the last term on the right - hand side of ( 6.9 ) requires new calculations because it contains functional derivatives of @xmath250 .
these can be dealt with after taking infinitesimal variations of an equation like ( 2.4 ) . with the dewitt notation used here , the defining equation for the parametrix reads as @xmath275 where the signs on the right - hand side have been chosen so as to agree with eq .
( 1.4 ) when @xmath276 ( which means that the parametrix is actually a fundamental solution or even a green function ) .
the operator @xmath277 is obtained , in classical theory , as @xcite @xmath278 where @xmath279 is a local and symmetric matrix which is taken to transform like @xmath280 under group transformations , and @xmath281 is a local , nonsingular , symmetric matrix which transforms according to the adjoint representation of the infinite - dimensional invariance group .
such matrices act as metrics that raise and lower indices of the generators @xmath282 according to the rules @xcite @xmath283 if we study the transformation properties of eq .
( a3 ) under the infinitesimal gauge transformations ( a3 ) , we find @xmath284 now we act with @xmath285 on both sides of eq .
( 7.4 ) , and we find , by virtue of ( 7.1 ) , the relation @xmath286 at this stage we revert to condensed - index notation , hence writing , from ( 7.5 ) , @xmath287 where , in light of ( 7.2 ) , we can write @xmath288 at this stage , we are going to re - express in ( 7.7 ) the terms @xmath289 and @xmath290 .
for this purpose , inspired by the method adopted in ref .
@xcite , we begin by noticing that , for background fields that solve the classical field equations , one has ( see eq .
( a7 ) and set @xmath291 therein ) @xmath292 where the operator @xmath293 is defined by @xcite @xmath294 the associated parametrix @xmath295 is here defined , by analogy with ( 7.1 ) , in the form @xmath296 now we act first with @xmath297 on @xmath298 , finding , in light of ( 7.1 ) , the functional equation @xmath299 furthermore , we act upon eq .
( 7.11 ) with the parametrix @xmath295 and find , by virtue of ( 7.10 ) , the functional equation resulting from ( 7.11 ) , i.e. @xmath300 the left - hand side of ( 7.12 ) is exactly of the type @xmath290 occurring in eq .
if we were dealing with green functions , eq .
( 7.12 ) would reduce to the familiar relation @xcite @xmath301 interestingly , on going from green functions to parametrices , a basic functional equation like ( 7.13 ) receives terms linear in @xmath302 and @xmath303 . as a second step , we act with @xmath304 from the left on eq .
( 7.8 ) , finding that @xmath305 which implies @xmath306 this is the transposed of eq . ( 7.12 ) , and generalizes the more familiar functional equation for green functions @xcite @xmath307 by virtue of the group invariance property satisfied by all physical observables ( set @xmath308 or @xmath1 or @xmath221 in eq . ( a5 ) ) , the terms like @xmath309 and @xmath310 on the right - hand side of ( 7.12 ) and ( 7.15 ) give vanishing contribution to @xmath268 .
one is therefore left with the contributions involving third functional derivatives of the action @xmath17 , plus terms linear in @xmath302 and @xmath303 , hereafter denoted by @xmath311 .
bearing in mind that @xmath312,one can relabel indices summed over , finding eventually ( cf . ref .
@xcite ) @xmath313s_{,abc } \nonumber \\ & = & a_{,i}b_{,j}c_{,k}\bigr[(\pi_{+}^{ia}-\pi_{-}^{ia } ) ( \pi_{+}^{jb}\pi_{-}^{kc}-\pi_{-}^{jb}\pi_{+}^{kc } ) \nonumber \\ & + & ( \pi_{+}^{jb}-\pi_{-}^{jb})(\pi_{+}^{kc}\pi_{-}^{ia } -\pi_{-}^{kc}\pi_{+}^{ia } ) \nonumber \\ & + & ( \pi_{+}^{kc}-\pi_{-}^{kc})(\pi_{+}^{ia}\pi_{-}^{jb } -\pi_{-}^{ia}\pi_{+}^{jb})\bigr]s_{,abc}=0 , \label{(7.17)}\end{aligned}\ ] ] where we have assumed the nontrivial properties @xmath314 the sum in ( 7.17 ) vanishes because it involves six pairs of double products of parametrices with opposite signs , i.e. @xmath315
along the years , peierls brackets have been considered also in the modern literature to define poisson brackets on the space of histories @xcite , to define the feynman functional integral @xcite , to reconsider the covariant form of hamiltonian dynamics @xcite , to study poisson brackets for fermionic fields in de sitter space @xcite , and even to investigate the perturbative construction of models of algebraic quantum field theory @xcite .
the material in our secs .
ii and iii is necessary to achieve gradually the transition towards nonlinear hyperbolic equations , but at that early stage the parametrix is neither compelling nor more useful than the standard canonical approaches .
however , since the green functions of hyperbolic operators play such a key role in defining and evaluating peierls brackets , while for nonlinear hyperbolic equations not even the fundamental solution is defined , even nowadays , our approach comes into play .
we have exploited the work in ref .
@xcite to argue that , since one knows how to solve linear hyperbolic systems through integral equations , and how to obtain the solution of nonlinear hyperbolic systems from the solution of linear ones , one can obtain good computational recipes for approximate evaluation of peierls brackets , even though the exact peierls bracket for gravity , obtainable in principle from the fully covariant formalism , remains inaccessible .
more precisely , in the quantum commutator ( 1.15 ) for gravity , the factors on the right - hand side do not commute .
this means that we have to study ( 6.8 ) when @xmath4 is replaced by its unknown operator version , here denoted by boldface characters , i.e. @xmath316 , for which @xmath317 @xmath318 at this stage , our idea is to generalize the splits ( 6.1 ) , ( 6.2 ) and the definitions ( 6.3 ) , _ assuming _ that the operator green s function @xmath319 has a classical part given by the parametrix @xmath246 and a quantum part given by the as yet unknown operator @xmath320 , the latter being responsible for the lack of commutativity in ( 8.1 ) and ( 8.2 ) .
hence we write @xmath321 it is this operator equation that should be inserted into the operator version of @xmath266 and @xmath322 in ( 6.8 ) , and this is the technical challenge ahead of us in the years to come . in our paper
we have developed ideas and have exploited well - established properties of hyperbolic and elliptic partial differential equations to obtain a novel perspective on the commutators of canonical quantum gravity .
we have shown that the fully diff - invariant classical peierls bracket among diff - invariant functionals in general relativity can be approximated by means of a parametrix of a fifth - order elliptic operator , convoluted with functions solving kirchhoff - type formulas .
the procedure is more systematic than the eq .
( 1.14 ) leading to the advanced and retarded green functions , which was advocated but not solved in ref .
although the parametrix of a partial differential operator is not unique , this apparent drawback can be exploited in order to evaluate with increasing accuracy the desired classical peierls bracket .
after all , the best we can do in science is to evaluate with high accuracy the functional relations we are interested in .
we have asumed that the operator green functions of canonical quantum gravity consist of a classical part given by the above parametrix , plus a quantum part that is responsible of technical complications but should be kept under control . our original work in secs from v to vii puts on firm ground the existence of the desired parametrix for gravity , and derives the novel functional equations ( 7.6 ) , ( 7.7 ) , ( 7.12 ) and ( 7.15 ) that are a concrete step towards applying parametrices in canonical quantization of field theories possessing an infinite - dimensional invariance group . from the point of view of general formalism ,
another interesting issue is whether a relation exists with the work in ref .
@xcite , where the author has studied a formulation of the ordering problems of general relativity inspired by the grnewald - van hove theorem . this work of segre establishes a negative result , i.e. , the lack of a suitable quantization map and of a suitable extension of such a map to general relativity .
we hope to understand better the issue in future , so as to investigate the possible implications ( if any ) for the quantum part @xmath320 of the green function in our eq .
( 8.3 ) .
if it were possible to achieve our goals , a novel perspective on the longstanding problems of ( canonical ) quantum gravity would emerge .
g. e. is grateful to the dipartimento di fisica ettore pancini of federico ii university for hospitality and support , and to b. booss - bavnbek for correspondence . along the years
, the author has benefited from discussions and collaboration with g. bimonte , g. gionti , g. marmo and c. stornaiolo on the hamiltonian formalism in field theory .
the dewitt notation @xcite makes it possible to use matrix - like notation for the otherwise rather lengthy functional equations encountered in field theory .
it does not solve , by itself , the open problems of theoretical physics , but makes it possible to see easily what features are shared by seemingly very different field theories .
the field variables are denoted by @xmath323 , e.g. the electromagnetic potential @xmath163 , or the yang - mills potential @xmath324 or the spacetime metric @xmath325 . for each functional @xmath126 of the fields ,
its functional derivatives are denoted by @xmath326 @xmath327 and so on .
since the latin indices used here for fields carry information about the spacetime point where the field is evaluated , an expression like @xmath328 is meant to be @xmath329 unlike tensor calculus , where the einstein convention @xmath330 in @xmath197-dimensional spacetime is a short - hand notation for a sum without integration , i.e. @xmath331 . in the gauge theories of theoretical physics , the action functional remains unchanged in value under certain continuous changes in the dynamical variables , which are confined to finite but otherwise arbitrary spacetime domains .
the set of all such changes constitutes a transformation group .
the abstract group of which the transformation group forms a representation is the _ invariance group _ of the system @xcite . since the finite domain over which the changes in the dynamical variables differ from zero is arbitrary , the invariance group is necessarily infinite - dimensional , i.e. it is a pseudo - group @xcite . under an infinitesimal group transformation
the changes may be expressed in the general form @xmath332 where , for local theories , the @xmath333 are linear combinations of dirac s delta and its derivatives , while the @xmath334 are arbitrary infinitesimal differentiable functions of the spacetime points , known as group parameters , and vanish outside the arbitrary domain under consideration @xcite . when eq .
( a3 ) holds , the infinitesimal variation of any functional @xmath126 of the fields is expressed by @xmath335 in particular , if @xmath126 is gauge - invariant , its variation @xmath336 in ( a4 ) vanishes , and one finds @xmath337 where @xmath338 are vector fields on the space of field configurations , i.e. @xmath339 note also that , if @xmath126 coincides with the action functional @xmath17 of a gauge theory , functional differentiation of eq .
( a5 ) yields
@xmath340 thus , restriction of eq .
( a7 ) to the dynamical subspace , where the euler - lagrange equations @xmath291 hold , tells us that the operator @xmath280 is not invertible , because it then possesses nonvanishing eigenvectors belonging to the zero eigenvalue .
thus , one has to resort to the imposition of suitable supplementary conditions whenever the field theory is a gauge theory ( see , for example , eq .
( 1.11 ) taken from ref .
@xcite ) . for einstein s general relativity , the explicit form of @xmath280
was first obtained by levi - civita @xcite and then , in a more accessible reference , by dewitt @xcite , but it is remarkable that , before any detailed calculation , the lack of green functions or fundamental solutions of @xmath280 for any gauge theory is already clear from first principles .
a connected open set @xmath181 of the spacetime manifold @xmath169 is said to be a geodesically convex domain if any two points @xmath41 and @xmath129 in @xmath181 are joined by a unique geodesic in @xmath181 . in a geodesically convex domain @xmath39 ,
let the local coordinates of two points @xmath129 and @xmath41 be @xmath31 and @xmath341 , respectively .
if @xmath342 $ ] , is a parametrized @xmath343 curve joining @xmath41 and @xmath129 in @xmath39 , the arc - length of @xmath344 is defined by @xmath345 when @xmath344 is the unique geodesic joining @xmath41 and @xmath129 in @xmath39 , the arc - length @xmath346 is said to be the _ geodesic distance _ of @xmath41 and @xmath129 , and the square of the geodesic distance is @xmath347 : @xmath348 and is written as @xmath349 in local coordinates . in the literature on general relativity , @xmath347 is called the world function .
it was first introduced by hadamard @xcite in the analysis of the cauchy problem for linear partial differential equations , then by ruse @xcite and synge @xcite , and eventually by dewitt and brehme @xcite in the investigation of electromagnetic green functions on curved spacetime . in a geodesically convex domain @xmath39 , the future ( resp .
past ) null semi - cone @xmath350 ( resp .
@xmath351 ) is the set of all points @xmath129 of @xmath39 such that there exists a future - directed ( resp .
past - directed ) null geodesic from @xmath41 to @xmath129 . the future ( resp .
past ) domain of dependence @xmath352 ( resp .
@xmath353 ) is the set of all points @xmath354 that can be reached along future - directed ( resp . past - directed ) timelike geodesics from @xmath41 .
one has therefore @xmath355 the closure of @xmath352 ( resp .
@xmath353 ) is instead called the future ( resp .
past ) emission of @xmath41 : @xmath356 such sets are also called causal future ( @xmath357 ) and causal past ( @xmath358 ) of @xmath41 @xcite . in eqs .
( 2.12 ) and ( 2.13 ) , we have @xmath359 moreover , @xmath360 are distributions that act on test functions @xmath361 .
a connected open set @xmath39 is called a _ causal domain _ if 0.3 cm ( i ) there is a geodesically convex domain @xmath120 such that @xmath362 , and 0.3 cm ( ii ) @xmath363 , the set @xmath364 is either a compact subset of @xmath39 , or otherwise is empty .
for systems of linear second - order partial differential equations , the theory of characteristics is developed starting from the equations @xcite @xmath365 which are soluble with respect to @xmath366 ( this is the _ normality _ condition ) if @xmath367 .
we then ask ourselves under which conditions is the normality character preserved if , to the independent variables @xmath368 , we apply the transformation @xmath369 so that the hyperplane @xmath370 is transformed into an hypersurface having equation @xmath371 starting from which one can determine , at least in a neighborhood , the @xmath372 functions . for this purpose ,
let us consider the covector ( or covariant vector or one - form ) variables @xmath373 from which we get @xcite @xmath374 @xmath375 hence the original system ( c1 ) gets transformed into @xmath376 thus , upon considering the principal ( or leading ) symbol of the operator in eqs .
( c1 ) , i.e. the matrix of polynomials @xmath377 also called the characteristic polynomial , the normality condition of ( c1 ) is preserved if @xmath378 . if instead @xmath39 vanishes , it is no longer possible to apply ( whatever the value of @xmath379 ) the cauchy theorem starting from the hypersurfaces @xmath380 .
the equation @xmath381 is the equation defining the _ characteristics manifolds _ @xcite . in particular , the characteristic surfaces of the system ( 4.10 ) are three - dimensional manifolds ( null hypersurfaces @xcite ) of four - dimensional spacetime with coordinates @xmath190 and solve the differential system @xmath382 @xmath383 the four quantities @xmath384 denote a system of directional parameters of the normal vector field .
let us take this system , which is only defined up to a proportionality factor , in such a way that @xmath385 , and let us set @xmath386 .
the desired surfaces are therefore a solution of @xmath387 @xmath388 the characteristics of this differential system , which are bicharacteristics ( i.e. null geodesics @xcite ) of the eqs .
( 4.10 ) , satisfy the following differential equations : @xmath389 @xmath390 being an auxiliary parameter .
the characteristic conoid @xmath391 with vertex @xmath392 is the characteristic surface generated from the bicharacteristics passing through @xmath393 .
such a bicharacteristic solves the system of integral equations @xcite @xmath394 @xmath395 @xmath396 where the @xmath397 satisfy the relation ( c3 ) , i.e. @xmath398 @xmath399 being the value of @xmath187 at the vertex @xmath393 of the conoid @xmath391 .
one can assume the following values : @xmath400 so that ( c17 ) reduces to @xmath401 besides the parameter @xmath390 which defines the position of a point on a given bicharacteristic , we will need two more parameters @xmath402 and @xmath403 which vary with the bicharacteristic under investigation and are given by @xcite @xmath404 moreover , following again ref .
@xcite , for any function @xmath405 of the spacetime coordinates @xmath190 , we denote by @xmath406 $ ] its restriction to the characteristic conoid , i.e. @xmath407 \equiv \left .
\varphi(x^{\alpha } ) \right |_{\sigma_{0 } } , \label{(c21)}\ ] ] and we consider the second - order operator @xmath408 { \partial^{2}\varphi \over \partial x^{i } \partial x^{j } } , \label{(c22)}\ ] ] and its ( formal ) adjoint @xmath409\psi ) \over \partial x^{i } \partial x^{j}}. \label{(c23)}\ ] ] at this stage , the work in ref .
@xcite took linear combinations of eqs .
( 4.10 ) , with left - hand side ( with free index @xmath410 ) denoted by @xmath411 , the coefficients of linear combination being some a priori unknown functions @xmath412 .
remarkably , such functions were found to solve a set of equations and turned out to obey the factorization property @xmath413 where @xmath414 having defined @xmath415+[b_{t}^{ri}]\xi_{i}\right ) , \label{(c26)}\ ] ] @xmath416\right ) -{1\over
2}\sum_{i=1}^{3}\left({\partial \over \partial x^{i}}[a^{i0}]\right ) , \label{(c27)}\ ] ] while , on denoting by @xmath417 the jacobian of the change of variables @xmath418 on the conoid @xmath391 , one finds @xcite @xmath419 the work in ref .
@xcite arrived at kirchhoff formulas for the solution of eqs .
( 4.10 ) where the integrand is built from the functions @xmath420\sigma_{s}^{r}{\partial [ u_{r}]\over \partial x^{j } } -[u_{r}]{\partial \over \partial x^{j}}([a^{ij}]\sigma_{s}^{r } ) + [ b_{ri}^{t}][u_{t}]\sigma_{s}^{r } \nonumber \\ & + & 2 \sigma_{s}^{r}\left \ { [ a^{ij}]\xi_{j}+[a^{i0 } ] \right \ } \left[{\partial u_{r}\over \partial x^{0}}\right ] \biggr \ } , \label{(c30)}\end{aligned}\ ] ] @xmath421\sigma_{s}^{t}\right ) , \label{(c31)}\ ] ] and also from the parameter @xmath126 resulting from the following geometric considerations . the surfaces upon which we perform integration are surfaces @xmath422 traced on the characteristic conoid @xmath391 . thus , in light of ( c10 ) , they fulfill the differential relation @xmath423 . to compute the surface element @xmath424 we re - express the volume element , originally written as @xmath425 by exploiting the surfaces @xmath422 and the bicharacteristics ( where only @xmath390 is varying ) ,
i.e. @xcite @xmath426 where @xmath427 is the length element of the bicharacteristic , and @xmath428 is the angle of the bicharacteristic with the normal to the surface @xmath17 at the point considered . having defined all the concepts we need , we can now state the theorem proved in ref .
@xcite for linear hyperbolic systems .
0.3 cm * theorem*.
let the linear hyperbolic system ( 4.10 ) be given , satisfying the following assumptions : 0.3 cm ( i ) at the point @xmath393 , the conditions ( c18 ) hold . 0.3 cm ( ii ) the functions @xmath187 and @xmath429 have partial derivatives with respect to @xmath190 of order @xmath430 and @xmath431 , respectively , continuous and bounded in a domain @xmath432 the functions @xmath433 are continuous and bounded . 0.3 cm ( iii ) the fourth partial derivatives of @xmath187 and the second partial derivatives of @xmath429 fulfill lipschitz conditions .
0.3 cm then every continuous , bounded solution of the eqs .
( 4.10 ) with continuous and bounded first derivatives within @xmath181 satisfies the kirchhoff integral relations @xmath434l_{s}^{r}+\sigma_{s}^{r}[f_{r}]\right ) { j \over t^{0 } } { \rm d}x^{0 } \ ; { \rm d}\lambda_{2 } \ ; { \rm d}\lambda_{3 } \nonumber \\ & + & \int_{0}^{2 \pi } \int_{0}^{\pi } \left \{{e_{s}^{r}j \xi_{r}\over t^{0 } } \right \}_{x^{0}=0 } { \rm d}\lambda_{2 } \ ; { \rm d}\lambda_{3 } \biggr ] , \label{(c33)}\end{aligned}\ ] ] if the coordinates @xmath435 of @xmath393 fulfill majorizations of the form @xmath436 which define a domain @xmath437 .
a. m. dirac , the fundamental equations of quantum mechanics , proc .
lond . a * 109 * , 642 ( 1926 ) . c. itzykson and j. b. zuber , _ quantum field theory _
( mcgraw - hill , new york , 1985 ) .
g. t hooft , _ 50 years of yang - mills theory _
( world scientific , singapore , 2005 ) .
a. m. dirac , the theory of gravitation in hamiltonian form , proc .
a * 246 * , 333 ( 1958 ) .
b. s. dewitt , quantum theory of gravity .
i. the canonical theory , phys . rev . * 160 * , 1113 ( 1967 ) .
j. a. wheeler , superspace and the nature of quantum geometrodynamics , in _ battelle rencontres _ , eds . c. m. dewitt and j. a. wheeler ( benjamin , new york , 1968 ) .
a. ashtekar , new variables for classical and quantum gravity , phys .
* 57 * , 2244 ( 1986 ) . c. rovelli , _ quantum gravity _ ( cambridge university press , cambridge , 2004 ) .
t. thiemann , _ modern canonical quantum general relativity _ ( cambridge university press , cambridge , 2008 ) . c. rovelli and f. vidotto , _ covariant loop quantum gravity _
( cambridge university press , cambridge , 2015 ) . y. foures - bruhat , existence theorem for certain systems of nonlinear partial differential equations , acta math . *
88 * , 141 ( 1952 ) .
r. arnowitt , s. deser , and c. w. misner , the dynamics of general relativity , gen .
rel . grav . * 40 * , 1997 ( 2008 ) .
y. choquet - bruhat , _ general relativity and the einstein equations _
( clarendon press , oxford , 2009 ) . c. j. isham and k. kuchar , representations of spacetime diffeomorphisms .
1 . canonical parametrized field theories , ann . phys .
( n.y . ) * 164 * , 288 ( 1985 ) . c. j. isham and k. kuchar , representations of spacetime diffeomorphisms .
canonical geometrodynamics , ann .
* 164 * , 316 ( 1985 ) .
g. esposito , g. gionti , and c. stornaiolo , spacetime covariant form of ashtekar s constraints , nuovo cimento b * 110 * , 1137 ( 1995 ) . g. esposito and c. stornaiolo , linear form of canonical gravity , nuovo cimento b * 111 * , 271 ( 1996 ) .
r. e. peierls , the commutation laws of relativistic field theories , proc .
a * 214 * , 143 ( 1952 ) .
b. s. dewitt , invariant commutators for the quantized gravitational field , phys .
lett . * 4 * , 317 ( 1960 ) .
b. s. dewitt , the spacetime approach to quantum field theory , in _ relativity , groups and topology ii _ ,
eds . b. s. dewitt and r. stora ( north - holland , amsterdam , 1984 ) . b. s. dewitt , _ the global approach to quantum field theory _ , int .
. phys . * 114 * , 1 ( clarendon press , oxford , 2003 ) .
p. a. m. dirac , _ the principles of quantum mechanics _
( clarendon press , oxford , 1958 ) .
f. g. friedlander , _ the wave equation on a curved space - time _ ( cambridge university press , cambridge , 1975 ) .
f. john , _ partial differential equations _ , applied mathematical sciences vol . 1 ( springer , berlin , 1975 ) .
p. r. garabedian , _ partial differential equations _ ( chelsea publishing , new york , 1964 ) .
f. treves , _ introduction to pseudodifferential and fourier integral operators .
volume 2 _ ( plenum press , new york , 1980 ) .
j. h. van vleck , the correspondence principle in the statistical interpretation of quantum mechanics , proc .
* 14 * , 178 ( 1928 ) . c. morette , on the definition and approximation of feynman s path integrals , phys .
rev . * 81 * , 848 ( 1951 ) .
l. lorenz , on the identity of the vibrations of light with electrical currents , phil . mag . * 34 * , 287 ( 1867 ) ; e. t. whittaker , _ the history of the theories of aether and electricity _ ( longman , london , 1910 ) .
l. hrmander , _ the analysis of linear partial differential operators .
( springer - verlag , berlin , 1983 ) .
g. bimonte , g. esposito , g. marmo , and c. stornaiolo , peierls brackets in field theory , int
. j. mod phys .
a * 18 * , 2033 ( 2003 ) .
b. s. dewitt , _ dynamical theory of groups and fields _ ( gordon & breach , new york , 1965 ) .
d. marolf , poisson brackets on the space of histories , ann .
( n.y . ) * 236 * , 374 ( 1994 ) .
d. marolf , the generalized peierls bracket , ann .
( n.y . ) * 236 * , 392 ( 1994 ) .
b. dewitt and c. dewitt - morette , from the peierls bracket to the feynman functional integral , ann .
( n.y . ) * 314 * , 448 ( 2004 ) .
h. ozaki , covariant description of hamiltonian form for field dynamics , ann .
( n.y . ) * 319 * , 364 ( 2005 ) .
g. esposito and r. roychowdhury , spinor two - point functions and peierls bracket in de sitter space , int .
methods mod .
* 6 * , 1069 ( 2009 ) .
k. fredenhagen and k. rejzner , perturbative construction of models of algebraic quantum field theory , arxiv:1503.07814 [ math - ph ] .
g. segre , a grnewald van hove like formulation of the ordering problems of general relativity , arxiv : math - ph/0512061 .
t. levi - civita , characteristics and bicharacteristics of the einstein gravitational equations , rend .
lincei ( sci .
fis . mat .
nat . ) * 11 * , 113 ( 1930 ) .
j. hadamard , _ lectures on cauchy s problem in linear partial differential equations _ ( dover , new york , 1952 ) .
h. s. ruse , taylor s theorem in the tensor calculus , proc .
london math .
soc . * 32 * , 87 ( 1931 ) .
j. l. synge , a characteristic function in riemannian space and its application to the solution of geodesic triangles , proc .
london math .
soc . * 32 * , 241 ( 1931 ) .
b. s. dewitt and r. w. brehme , radiation damping in a gravitational field , ann .
* 9 * , 220 ( 1960 ) .
s. w. hawking and g. f. r. ellis , _ the large - scale structure of space - time _
( cambridge university press , cambridge , 1973 ) . t. levi - civita , _ characteristics of differential systems and wavelike propagation _ ( zanichelli , bologna , 1931 ) .
|
in the sixties , dewitt discovered that the advanced and retarded green functions of the wave operator on metric perturbations in the de donder gauge make it possible to define classical poisson brackets on the space of functionals that are invariant under the action of the full diffeomorphism group of spacetime .
he therefore tried to exploit this property to define invariant commutators for the quantized gravitational field , but the operator counterpart of such classical poisson brackets turned out to be a hard task . on the other hand , in the mathematical literature , it is by now clear that , rather than inverting exactly an hyperbolic ( or elliptic ) operator , it is more convenient to build a quasi - inverse , i.e. an inverse operator up to an operator of lower order which plays the role of regularizing operator .
this approximate inverse , the parametrix , which is , strictly , a distribution , makes it possible to solve inhomogeneous hyperbolic ( or elliptic ) equations .
we here suggest that such a construction might be exploited in canonical quantum gravity provided one understands what is the counterpart of classical smoothing operators in the quantization procedure .
we begin with the simplest case , i.e. fundamental solution and parametrix for the linear , scalar wave operator ; the next step are tensor wave equations , again for linear theory , e.g. maxwell theory in curved spacetime .
last , the nonlinear einstein equations are studied , relying upon the well - established choquet - bruhat construction , according to which the fifth derivatives of solutions of a nonlinear hyperbolic system solve a linear hyperbolic system .
the latter is solved by means of kirchhoff - type formulas , while the former fifth - order equations can be solved by means of well - established parametrix techniques for elliptic operators .
but then the metric components that solve the vacuum einstein equations can be obtained by convolution of such a parametrix with kirchhoff - type formulas . some basic functional equations for the parametrix
are also obtained , that help in studying classical and quantum version of the jacobi identity .
|
irreducible nonsingular @xmath0-matrices are one class of the most important matrices from applications , such as discretized pdes , markov chains abate94,rober90 and electric circuits @xcite , and have been studied extensively in the literature ; see , for instance , ( * ? ? ?
* chapter 6 ) . in many applications ,
one is interested in finding the smallest eigenvalue @xmath2 and the associated eigenvector @xmath3 of an irreducible nonsingular @xmath0-matrix @xmath4 .
@xmath0-matrices are closely related to nonnegative matrices .
for instance , an @xmath5-matrix @xmath6 can be expressed in the form @xmath7 with a nonnegative matrix @xmath8 and some constant @xmath9 , the spectral radius of @xmath10 ; @xmath11 is nonnegative . for more properties and a systematic account of @xmath0-matrices and nonnegative matrices ,
see @xcite .
nonnegative matrices have important applications in many areas @xcite , including economics , statistics and network theory . for a nonnegative matrix @xmath10 ,
at least one of the eigenvalues of maximal magnitude is nonnegative and hence equal to the spectral radius @xmath12 .
the corresponding eigenvectors @xmath3 satisfy @xmath13 and are called the perron vectors of @xmath10 if they are nonnegative .
the nonnegative @xmath10 always has at least one perron vector . in applications ,
the perron vectors play an important role , and they describe , e.g. , an equilibrium , a probability distribution or an optimal network property @xcite . furthermore ,
if @xmath10 is irreducible , then the perron - frobenuis theorem hjo85 states that @xmath14 is simple and there is a positive eigenvector @xmath3 associated with @xmath14 .
one is often interested in verifying the uniqueness and strict positivity of the perron vector of the irreducible nonnegative matrix @xmath10 @xcite .
the well - known pagerank vectors are special perron vectors of very large google matrices whose largest eigenvalues are equal to one @xcite . from the relation @xmath7 , we see that the smallest eigenvalue @xmath15 of the irreducible nonsingular @xmath0-matrix @xmath6 is simple and equal to @xmath16 , and the positive vector @xmath3 is the unique associated eigenvector up to scaling . consequently , if @xmath17 is available , then the computation of the smallest eigenpair @xmath18 of @xmath6 amounts to that of the largest eigenpair of @xmath10 . for a general large and sparse @xmath6
, there are a number of general numerical methods for computing a small number of its eigenpairs .
krylov type methods including the power method applied to @xmath6 directly are suitable for exterior eigenvalues , i.e. , some eigenvalues close to the exterior of the spectrum , and the associated eigenvectors , but they face serious challenges when the desired eigenvalues are in the interior of the spectrum .
many methods , such as inverse iteration , rayleigh quotient iteration ( rqi ) and shift - invert arnoldi , have been developed to overcome these difficulties ; see saa92,par98,ste01 .
however , they require the solution of a possibly ill - conditioned large linear system involving a shifted @xmath6 , called inner linear system , at each iteration .
this is generally very difficult and even impractical by a direct solver since a factorization of a shifted @xmath6 may be expensive or prohibited .
when inner linear systems are approximately solved by iterative solvers , we are led to inner - outer iteration methods , also called inexact eigensolvers .
the inner iteration means that the inner linear system at each step is approximately solved iteratively , while the outer iteration is the update of the approximate eigenpair(s ) . throughout this paper
, we always make the underlying hypothesis that a direct solver is not viable for large sparse linear systems and only iterative solvers can be used . there has been ever growing and intensive interest in inexact eigensolvers over last two decades . among them , inexact inverse iterationlll97,bgs06,liu12 and inexact rqi @xcite are the simplest and the most basic ones .
in addition , they are often key ingredients of other sophisticated and practical inexact methods , such as inverse subspace iteration @xcite , the jacobi davidson method svo96,ste01 and the shift - invert residual arnoldi method @xcite . in the mentioned papers and the references therein as well as some others , a number of theoretical results have been established on these methods .
particularly , it has been shown @xcite that , different from rqi , the inexact rqi generally does not converges globally , rather it is only locally convergent by requiring that the initial vector is a reasonably good approximate eigenvector .
because of these two reasons , the inexact rqi is seldom used in practice unless a good initial guess to the desired eigenvector is already available . for the computation of the perron vector of the nonnegative @xmath10 and the eigenvector of the @xmath0-matrix @xmath6 associated with the smallest eigenvalue ,
a central concern is how to preserve the strict positivity of approximate eigenvectors . for nonnegative matrices and @xmath0-matrices ,
such positivity is crucial in some applications since if all the components of an approximate eigenvector do not have the same sign then it may be physically meaningless and could not be interpreted .
unfortunately , all the methods but the power method mentioned previously are not structure preserving and can not guarantee the desirable positivity of approximations since it is possible that a converged approximation of @xmath3 may well have negative components , as is the typical case when the unit length @xmath3 has very small components .
theoretically , the power method fits into this purpose for nonnegative matrices and naturally preserves the strict positivity of approximate eigenvectors , provided that the starting vector is positive .
due to the equivalence of nonnegative matrix and @xmath0-matrix eigenproblems , the power method can be adapted to computing the smallest eigenpair of an irreducible nonsingular @xmath0-matrix .
however , the power method , though globally convergent , is generally very slow and may be impractical .
therefore , it is very appealing in both theory and practice to develop both efficient and reliable positivity preserving numerical methods for the nonnegative matrix and @xmath0-matrix eigenproblems .
we will devote ourselves to this topic in this paper .
in @xmath19 , noda @xcite introduced an inverse iteration method with variable shifts for nonnegative matrix eigenvalue problems .
this iteration method is a structure preserving method and was motivated by the works of collatz @xcite in 1942 and wielandt @xcite in 1950 ; see ( * ? ? ? * and 59 ) ,
@xcite and @xcite for a description and historic overview .
however , it was noda who proposed its inverse iteration form with the variable shifts different from rayleigh quotients .
we , therefore , call the iteration the noda iteration ( ni ) . given a positive starting vector ,
ni naturally preserves the strict positivity of approximate eigenvectors at all iterations .
it has been adapted to computing the smallest eigenpair of an irreducible nonsingular @xmath0-matrix xue96,axy02 .
the purpose of @xcite is to compute the smallest eigenvalue of such @xmath0-matrix with high relative accuracy .
there , it has been shown that ni is practical and effective for such a pursue , in which the linear systems involved are solved accurately by a special direct solver , called the gth like algorithm .
it is well known that rqi is almost always globally convergent for any starting vector ( * ? ? ?
* theorem 4.9.1 ) , but its correct convergence to a desired eigenpair is conditional and requires that the initial vector be a reasonably accurate approximation to the desired eigenvector @xcite . as a result ,
rqi itself is seldom used to solve a practical problem unless a good initial guess is available .
in contrast , a major advantage of ni is that , for any positive initial vector , it converges globally and computes the desired eigenpair correctly .
furthermore , the convergence order of ni is asymptotically superlinear @xcite and actually quadratic @xcite .
as it will become clear , ni always generates a monotonically decreasing sequence of approximate eigenvalues whose convergence to @xmath14 is guaranteed ; for the smallest eigenvalue @xmath2 of the irreducible nonsingular @xmath0-matrix @xmath6 , it always generates a monotonically increasing sequence of approximate eigenvalues that converge to @xmath2 unconditionally . in other words ,
the approximate eigenvalues converge to @xmath14 from above or converge to the smallest eigenvalue of the @xmath0-matrix from below .
in contrast , for symmetric matrices , rayleigh quotients always reside in the spectrum interval . as a result ,
if rqi converges correctly , the sequence of approximations approach the perron root of the nonnegative matrix and the smallest eigenvalue of the @xmath0-matrix from inside to outside . in this paper , keep in mind that only iterative solvers are supposed to be viable to solve inner linear systems approximately .
based on ni , we first propose an inexact noda iteration ( ini ) to find the perron root and vector of an irreducible nonnegative matrix @xmath10 . as an inexact eigensolver ,
our major contribution is to propose two practical inner tolerance strategies for solving the inner linear systems involved , so that the resulting two ini algorithms are structure preserving and globally converge .
the first inner tolerance strategy uses @xmath20 as a stopping criterion for inner iterations with the constant @xmath21 and @xmath22 is the current positive approximate eigenvector .
the second inner tolerance strategy solves the inner linear systems with certain decreasing tolerances for inner iterations , which will be described in the context .
we establish a rigorous convergence theory of ini with these two inner tolerance strategies , proving that the convergence of the former iteration is globally linear with the asymptotic convergence factor bounded by @xmath23 and that of the latter is asymptotically superlinear with the convergence order @xmath1 , respectively . in order to derive this superlinear convergence order
, we establish a close relationship between the eigenvalue error and the eigenvector error obtained by ni and ini , which is interesting in its own right .
we also revisit the convergence of ni and establish a new quadratic convergence result different from that in @xcite .
as we will see , the ini algorithms developed and the theory established are easily extended to the computation of the smallest eigenpair of an irreducible nonsingular @xmath0-matrix @xmath6 .
finally , we stress that , different from @xcite , our aim is the positivity preserving computation of the desired eigenvectors , while their concern is the relative high accuracy computation of the perron root and the smallest eigenvalue of an @xmath0-matrix .
the rest of this paper is organized as follows . in section 2
, we introduce ni and some preliminaries . in section 3 ,
we present an inexact noda iteration and prove some basic properties of it . in section 4 , we propose two practical ini algorithms , called ini_1 and ini_2 , respectively , for computing the spectral radius and the perron vector of an irreducible nonnegative matrix .
we then establish their global convergence theory .
moreover , we precisely derive the asymptotic linear convergence factor of ini_1 and superlinear convergence order of ini_2 . in section 5 , we adapt ini_1 and ini_2 to the computation of the smallest eigenvalue and the corresponding vector of an irreducible nonsingular @xmath0-matrix and present their convergence theory . in section 6 , we report the numerical results on a few practical problems to justify the convergence theory of ini and illustrate their effectiveness .
we also compare ini with the jacobi davidson method @xcite , the implicitly restarted arnoldi method @xcite and the explicitly restarted krylov
schur method ste02 , all of which are not positivity preserving .
the experiments indicate that the proposed ini algorithms always preserve the positivity of approximate eigenvectors , while the other three methods often fail to do so .
also , we demonstrate that the ini algorithms are efficient , competitive with and can outperform the other three methods considerably . finally , we summarize the paper with some concluding remarks in section 7 .
for any real matrix @xmath24 \in \mathbb{r}^{n\times n}$ ] , we denote @xmath25 $ ] .
if the entries of @xmath10 are all nonnegative ( positive ) , then we write @xmath8 @xmath26 . for real matrices @xmath10 and @xmath27 of the same size ,
if @xmath28 is a nonnegative matrix , we write @xmath29 . a nonnegative (
positive ) vector is similarly defined .
a nonnegative matrix @xmath10 is said to be reducible if there exists a permutation matrix @xmath30 such that@xmath31,\ ] ] where @xmath32 and @xmath33 are square matrices ; otherwise it is irreducible . here
the superscript @xmath34 denotes the transpose of a vector or matrix . throughout the paper
, we use a @xmath35-norm for vectors and matrices , and all matrices are @xmath36 unless specified otherwise .
we review some fundamental properties of nonnegative matrices and @xmath0-matrices .
[ thm : m ] let @xmath37 .
then the following statements are equivalent : 1 .
@xmath38 , @xmath39 for @xmath40 , and @xmath41 ; 2 .
@xmath7 with some @xmath42 and @xmath43 .
matrices having the above properties are called nonsingular @xmath0-matrices . for a pair of positive vectors
@xmath44 and @xmath45 , define @xmath46 where @xmath47^t$ ] and @xmath48^t$ ] .
the following lemma gives bounds for the spectral radius of a nonnegative matrix @xmath10 ; see @xcite .
[ maxmin]let @xmath49be an irreducible nonnegative matrix .
if @xmath50 is not an eigenvector of @xmath10 , then @xmath51 suppose that @xmath6 is an irreducible nonsingular @xmath0-matrix and @xmath18 is the smallest eigenpair of it
. then if @xmath52 is not an eigenvector of @xmath6 , it is easily justified from ( [ eq : maxmin ] ) and @xmath7 that @xmath53 for an irreducible nonnegative matrix @xmath10 , recall that the largest eigenvalue @xmath14 of @xmath10 is simple .
let @xmath3 be the unit length positive eigenvector corresponding to @xmath14 .
then for any orthogonal matrix @xmath54 $ ] it holds ( cf .
@xcite ) that @xmath55 b\left [ \begin{array}{cc } \mathbf{x } & v\end{array}\right ] = \left [ \begin{array}{cc } \rho(b ) & \mathbf{c}^t \\ 0 & l\end{array}\right ] \label{eqn : partition}\]]with @xmath56 whose eigenvalues constitute the other eigenvalues of @xmath10 . if @xmath57 is not an eigenvalue of @xmath58 , the sep function for @xmath57 and @xmath58 is defined as @xmath59 which is well defined as @xmath60 since @xmath14 is simple . throughout the paper , we will denote by @xmath61 the acute angle of any two nonzero vectors @xmath45 and @xmath62 . in @xcite
, noda presented an inverse iteration with variable shifts for computing the perron root and vector of an irreducible nonnegative matrix @xmath10 .
given an initial guess @xmath63 with @xmath64 , the noda iteration ( ni ) is an inverse iteration with variable shifts , and each iteration consists of three steps @xmath65the main step is to compute a new approximation @xmath66 to @xmath3 by solving the inner linear system ( [ eq : step1 ] ) .
lemma maxmin shows that @xmath67 @xmath14 as long as @xmath22 is not a scalar multiple of eigenvector @xmath3 .
furthermore , since @xmath68 is an irreducible nonsingular @xmath0-matrix , its inverse is irreducible nonnegative .
therefore , we have @xmath69 and @xmath70 , meaning that the above iteration scheme preserves the strict positivity of approximate eigenvector sequence @xmath71 .
we also see that ni is different from rqi , where , in the symmetric case , the rayleigh quotient of @xmath10 with respect to any vector lies in the spectrum interval of @xmath10 and thus no more than @xmath14 , and the approximate eigenvectors obtained by rqi do not preserve the strict positivity .
after variable transformation , we get the relation @xmath72 so @xmath73 is monotonically decreasing . based on ( eq : step1 ) ,
( [ eq : step2 ] ) and ( [ lambdaupdate ] ) , we can present ni as algorithm [ alg : iva ] . 1 .
given an initial guess @xmath74 with @xmath75 and @xmath76 , compute @xmath77 .
2 . * for * @xmath78 3 . solve the linear system @xmath79 .
4 . normalize the vector @xmath80 .
5 . compute @xmath81 .
* until * convergence : @xmath82 .
since it is supposed that only iterative solvers are viable to solve the linear system ( [ eq : step1 ] ) approximately at step 3 of algorithm alg : iva , in this section , we first propose an inexact noda iteration ( ini ) for the eigenvalue problem of an irreducible nonnegative matrix @xmath10 , and then we prove a number of basic properties of ini , which will be used to establish the convergence theory of two practical ini algorithms to be proposed in section 4 .
in ini we compute an approximate solution @xmath83 in step 3 of algorithm [ alg : iva ] , such that@xmath84where @xmath85 is the residual vector , whose norm @xmath86 is bounded by the inner tolerance at iteration @xmath87 .
[ monotone]let @xmath10 be an irreducible nonnegative matrix and @xmath88 be a fixed constant . for the unit length @xmath89 ,
if @xmath90 and @xmath85 in ( [ eq : inexactsys ] ) satisfies @xmath91then the new approximation @xmath92 in ( [ eq : normal ] ) and the sequence @xmath93 with @xmath94 is monotonically decreasing and bounded below by @xmath12 , i.e. , @xmath95 by @xmath96 and @xmath97 , it is known that @xmath98 . by lemma [ maxmin ] we know @xmath67 @xmath14 as @xmath99 .
consequently , @xmath100 is a nonsingular @xmath0-matrix , and the vector @xmath101 satisfies @xmath102 this implies @xmath103 and @xmath104 . from ( [ eq : inexactsys ] ) and the above it follows that@xmath105proving that he sequence @xmath93 is monotonically decreasing .
again , by lemma [ maxmin ] we have @xmath106 .
based on ( [ eq : inexactsys ] ) , ( [ eq : normal ] ) and ( [ eq : delamda ] ) and lemma [ monotone ] , we describe ini as algorithm [ alg : iiva ] . 1 . given an initial guess @xmath74 with @xmath75 , @xmath107 and @xmath76 , compute @xmath77 .
2 . * for * @xmath78 3 . solve @xmath79 approximately by an iterative solver such that @xmath108 4 .
normalize the vector @xmath109 .
5 . compute @xmath110 .
* until * convergence : @xmath82 .
note that if @xmath111 , i.e. , @xmath85 @xmath112 in ( [ eq : inexactsys ] ) for all @xmath87 then algorithm [ alg : iiva ] becomes the standard ni .
it follows from lemma [ monotone ] that algorithm [ alg : iiva ] generates the positive vector sequence @xmath71 , so it is a positivity preserving algorithm .
in what follows we will investigate convergence conditions of ini for @xmath113 as @xmath114 .
[ coslower]let @xmath115 be the unit length eigenvector of @xmath10 associated with @xmath12 . for any vector @xmath116 with @xmath117
, it holds that @xmath118 and @xmath119 we have @xmath120 . since @xmath115 and @xmath116 with @xmath121 , we have @xmath122 where @xmath123 is the vector 1-norm .
so ( [ eq : coslower ] ) holds .
we remark that the infimum in ( [ eq : coslower ] ) can not be attained because @xmath124 strictly for any @xmath116 with @xmath125 , but @xmath126 provided one of the components of @xmath62 tends to one and all the others tend to zero . by @xmath127
, it is easily seen that @xmath128 .
this upper bound is attained when all the components of @xmath3 are equal to @xmath129 . on the other hand
, we remark that lynn and timlake ( * ? ? ?
* theorem 2.1 ) derived a compact lower bound @xmath130 with the lower attained and equal to the upper bound @xmath131 when @xmath10 is a generalized stochastic matrix , i.e. , all the row sums of @xmath10 are equal to a positive constant . for such @xmath10 ,
its row sum is just @xmath14 .
so for @xmath132 , @xmath133 is always mildly small for @xmath134 large and can be very small if @xmath135 is very small .
let @xmath136 be generated by algorithm [ alg : iiva ] .
we decompose @xmath137 into the orthogonal direct sum@xmath138with @xmath139 and @xmath140 @xmath141 @xmath142 the acute angle between @xmath137 and @xmath3 .
now define @xmath143 then from ( [ eqn : partition ] ) we have @xmath144 b_{k}\left [ \begin{array}{cc } \mathbf{x } & v\end{array}\right ] = \left [ \begin{array}{cc } \varepsilon_{k } & \mathbf{c}^t \\ 0 & l_{k}\end{array}\right ] , \label{eqlpart2}\ ] ] where @xmath145 . for @xmath146 , it is easy to verify that @xmath144 b_{k}^{-1}\left [ \begin{array}{cc } \mathbf{x } & v\end{array}\right ] = \left [ \begin{array}{cc } \frac{1}{\varepsilon_{k } } & \mathbf{b}_k^t \\ 0 & l_{k}^{-1}\end{array}\right ] \mbox{\ \ with $ \mathbf{b}_k^t=-\frac{\mathbf{c}^tl_k^{-1}}{\varepsilon_{k}}$ } , \label{eqlpart3}\]]from which we get @xmath147 from lemma [ monotone ] , since @xmath148 is monotonically decreasing and bounded by @xmath14 from below , we must have @xmath149 , where @xmath150 or @xmath151 .
we next investigate these two possibilities , respectively , and present some basic results that will play an important role in proving @xmath150 when certain further restrictions are imposed on the inner tolerance @xmath152 .
[ equithm]let @xmath10 be an irreducible nonnegative matrix , and assume that @xmath153 is the largest eigenpair of @xmath10 with @xmath115 and @xmath154 .
if @xmath137 , @xmath155 , @xmath156 and @xmath85 are generated by algorithm [ alg : iiva ] , then the following statements are equivalent:@xmath157 ( i)@xmath158(ii)@xmath159 by the definition of @xmath160 , we get @xmath161 ( i)@xmath158(iii)@xmath159 since @xmath162 , from ( [ eq : decomposition ] ) we have@xmath163since @xmath164 , we can write @xmath165 with @xmath166 . from ( [ bkv ] ) and ( [ eq : decomyk ] ) , we get @xmath167 from lemma [ coslower ] it follows that @xmath168 for all @xmath87 . on the other hand , ( i )
means that @xmath169 and @xmath170 as @xmath171 .
since @xmath172 , we get @xmath173 . as a result , from definition ( [ eq : sep ] ) and @xmath174 , the second term in the right - hand side of ( [ decomp3 ] ) is bounded by@xmath175and @xmath176 combining ( [ eq : decomyk ] ) with the above and exploiting the norm triangle inequality , we have@xmath177 ( iii)@xmath158(ii)@xmath159 from @xmath178 with @xmath179 and @xmath180 we get @xmath181 meaning that @xmath182 since @xmath183 for @xmath171 , the sequence @xmath184 tends to a singular matrix , meaning that @xmath93 converges to an eigenvalue of @xmath10 . from lemma [ monotone ]
, we must have @xmath173 because @xmath93 is monotonically decreasing with the lower bound @xmath14 .
otherwise , @xmath185 would be an eigenvalue of @xmath10 bigger than the spectral radius @xmath14 of @xmath10 , which is impossible .
[ opposite ] for algorithm [ alg : iiva ] , if @xmath186 , then ( i ) @xmath187 is bounded@xmath188 ( ii ) @xmath189(iii ) @xmath190 with @xmath191 some constant .
\(i ) since @xmath97 , we get @xmath192(ii ) from ( [ eq : delamda ] ) it follows that @xmath193 on the other hand , from ( [ eq : bdyk ] ) and ( [ eq : minfk ] ) we have@xmath194 thus , we get@xmath195 ( iii ) suppose there is a subsequence @xmath196 that converges to zero .
then from lemma [ equithm ] there is a subsequence @xmath197 that converges to @xmath12 , a contradiction .
keep in mind that in lemmas [ equithm][opposite ] we only assume the condition @xmath198 with @xmath199 , under which we can only prove that the sequence @xmath200 converges to either @xmath14 or @xmath151 .
so only this condition is not enough to guarantee that ini computes the desired eigenpair @xmath201 of @xmath10 . in order to make @xmath202
, we have to impose some stronger conditions on @xmath203 .
in order to make ini converge correctly and effective , we now propose the following two practical inner tolerance strategies for the inexactness of step 3 of algorithm [ alg : iiva ] : * ini_1 : the residual norm satisfies @xmath204 for some fixed constant @xmath205 ; * ini_2 : the residual vector satisfies @xmath206 with @xmath207 for @xmath208 and @xmath209 for some constant @xmath205 .
it is easily seen that the residual vectors @xmath203 of ini_1 and ini_2 must satisfy @xmath210 with @xmath205 . as a result , it is known from lemma [ monotone ] that each of ini_1 and ini_2 generates a monotonically decreasing sequence @xmath211 bounded by @xmath14 and a sequence of positive vectors @xmath212 .
ini_1 and ini_2 now require stronger conditions on @xmath203 than the previous ini . in sections 4.14.2 , we will prove the global convergence of ini_1 and ini_2 , respectively .
furthermore , we will show that ini_1 converges at least linearly with the asymptotic convergence factor bounded by @xmath23 and ini_2 converges superlinearly with the asymptotic convergence order @xmath1 . in the meantime
, we revisit the convergence of ni and derive a new quadratic convergence result , which is included in section 4.2 .
we comment that since ini_2 converges , we must have @xmath213 , which means that we need to solve the linear systems more and more accurately as @xmath87 increases .
in contrast , the inner tolerance used by ini_1 is fixed except the factor @xmath214 , which tends to @xmath133 as @xmath171 whenever ini_1 .
this means that for a similar inner linear system , i.e. , @xmath73 almost the same , we may pay higher computational cost to solve it by ini_2 than ini_1 .
however , note that ini_1 and ini_2 converges linearly and superlinearly , respectively .
consequently , it is hard and should be impossible to make a general theoretical comparison of their overall efficiency .
actually , our numerical experiments will demonstrate that the overall efficiency of ini_1 and ini_2 is comparable , and there is no general winner between them .
we decompose @xmath66 in the same manner as ( [ eq : decomposition ] ) : @xmath215with @xmath216 .
so by definition , we have @xmath217 and @xmath218 .
obviously , @xmath219 if and only if @xmath220 , i.e. , @xmath221 . from ( [ eqlpart3 ] ) , we have @xmath222 exploiting the above relation , @xmath223 and @xmath224 , we obtain @xmath225 with the last second inequality holding by assuming that @xmath226 . this assumption must be satisfied provided that @xmath227 is suitably small , because , by lemma [ coslower ] , it holds that @xmath228 for all @xmath87 .
particularly , if @xmath229 , i.e. , @xmath230 , we recover ni and get @xmath231 we remark that @xmath232 if @xmath10 is a normal matrix as @xmath233 in this case . since ni is quadratically convergent @xcite , for @xmath87 large enough we must have @xmath234 therefore , for any given positive constant @xmath235
, it holds that @xmath236 for @xmath237 with @xmath238 large enough .
this means that , for @xmath237 , we have @xmath239 [ main1 ] let @xmath10 be an irreducible nonnegative matrix .
if the sequence @xmath93 is generated by ini_1 , then @xmath93 is monotonically decreasing and @xmath240 . by the assumption on ini_1 , since @xmath241 , it holds that @xmath97 with @xmath199 , which satisfies the condition in lemma monotone . so the sequence @xmath242 is bounded and monotonically decreasing , and we must have either @xmath243 or @xmath244 .
next we prove by contradiction that , for ini_1 , @xmath243 must hold .
suppose that @xmath245 .
by @xmath97 , we get @xmath246 it follows from ( ii ) of lemma [ opposite ] that @xmath247 thus , we have @xmath248from lemma [ coslower ] and ( iii ) of lemma [ opposite ] we know that @xmath249 and @xmath250 are uniformly bounded below by a positive constant . therefore , there is an @xmath251 such that @xmath252 and @xmath253 , leading to @xmath254using ( [ eq : tan ] ) and ( [ eq : lowersin ] ) , we obtain@xmath255define @xmath256 note that @xmath257 is uniformly bounded , and @xmath258 is a continuous function with respect to @xmath214 for @xmath259 .
then it holds that @xmath260 defined by ( [ niconv ] ) as @xmath261 .
particularly , for @xmath259 and any small positive number @xmath262 , it holds that @xmath263 provided that @xmath214 is suitably small . as a result , with @xmath264 defined by ( [ betak ] ) , for @xmath237
large enough we can choose a sufficiently small @xmath262 such that @xmath265 for @xmath214 sufficiently small .
as a result , for @xmath237 with @xmath238 large enough and @xmath214 sufficiently small , we have @xmath266 it then follows from this that @xmath267 , i.e. , @xmath268 . from lemma [ monotone ] , this means that @xmath93 converges to @xmath12 monotonically , a contradiction to the assumption that @xmath245 .
theorem [ main1 ] has proved the global convergence of ini_1 , but the result is only qualitative and does not tell us anything on how fast ini_1 converges .
next we precisely derive an upper bound for its asymptotic linear convergence factor . from ( [ eq : delamda ] ) and ( [ epsilonk ] ) , we have @xmath269with @xmath270 since @xmath271 , from ( [ rhok ] ) , ( [ eq : delamda ] ) and ( [ eq : monolam ] ) we always have @xmath272 [ linearconv]for ini_1 , we have @xmath273 , i.e. , the convergence of ini_1 is globally linear at least . since @xmath274 in ini_1
, it holds that @xmath97 .
therefore , we have @xmath275 as @xmath276 , it follows from the above relation that @xmath277 therefore , we have@xmath278from ( [ bkinv ] ) , we get @xmath279 since , from theorem [ main1 ] , @xmath280 and @xmath281 , we have @xmath282 and @xmath283 . on the other hand , since @xmath284 and @xmath285 , from ( [ rela3 ] ) we get @xmath286 consequently , we obtain @xmath287leading to @xmath288 it is seen from ( [ convfactor ] ) that if @xmath289 is small then ini_1 must ultimately converge fast . in this subsection
, we establish the global convergence theory of ini_2 and prove that its superlinear convergence order is @xmath1 .
in addition , we derive a relationship between the eigenvalue error @xmath290 and the eigenvector error @xmath291 , which holds for both ni and ini . in the meantime , as an important complement , we revisit the convergence of ni and prove its quadratic convergence in terms of @xmath291 .
[ main2]let @xmath10 be an irreducible nonnegative matrix . if @xmath160 is generated by ini_2 , then @xmath292 .
since @xmath293 with @xmath294 and @xmath209 with @xmath205 , @xmath203 satisfies the condition of lemma monotone .
therefore , the sequence @xmath200 generated by ini_2 is monotonically decreasing with the lower bound @xmath14 and converges to some limit @xmath295 .
this shows that @xmath296 as @xmath297 .
assume that @xmath298 .
then there exists a positive integer @xmath299 such that @xmath300 with @xmath289 defined in ini_1 . for @xmath262 , @xmath258 and @xmath238 in the proof of theorem [ main1 ] , take @xmath301 . then for @xmath302
we get @xmath303 hence , it holds that @xmath173 .
we further have the following result . [ eslonyk ] for ini_2
, it holds that @xmath304 where @xmath305 is defined by ( [ epsilonk ] ) . from ( [ eqlpart3 ] )
, we have @xmath306 therefore , we get @xmath307since @xmath283 and @xmath308 , we have @xmath309 , from which it follows that @xmath310 from lemma [ equithm ] and the proof of lemma [ main2 ] , we know that @xmath311 and @xmath312 , which lead to @xmath313 .
note that @xmath314 .
we then get @xmath315 a combination of ( [ rela1 ] ) and ( [ rela2 ] ) shows that @xmath316 [ superconv ] define the residual @xmath317 .
then for ini_2 , the following results hold : @xmath318that is , the convergence of ini_2 is superlinear .
( i ) : theorem [ main2 ] has proved that @xmath319 and @xmath320 . recall from the proof of theorem [ main2 ] that @xmath321 .
then it follows from ( eq : ratioerr ) and theorems [ main2][eslonyk ] that @xmath322(ii ) : from ( [ eq : ratioerr ] ) and ( [ eq:(i ) ] ) , we have@xmath323(iii ) : from ( [ eq : inexactsys ] ) , we have@xmath324from which it follows that@xmath325therefore , we get @xmath326from ( [ eq : lok ] ) , we have @xmath327 then it follows from theorem [ eslonyk ] , ( ii ) of theorem [ superconv ] and ( [ eq:(i)])([eq:(epslamd ) ] ) that @xmath328 although theorem [ superconv ] has established the superlinear convergence of ini_2 , it does not reveal the convergence order .
our next concern is to derive the precise convergence order of ini_2 .
this is more informative and instructive to understand how fast ini_2 converges .
elsner @xcite proved the ( asymptotic ) quadratic convergence of the sequence @xmath329 , but the constant factor ( multiplier ) in his quadratic convergence result appears hard to quantify or estimate .
below we establish an intimate and quantitative relationship between the eigenvalue error @xmath290 and the eigenvector error @xmath291 .
this result plays a crucial role in deriving the precise convergence order of ini_2 and proving the quadratic convergence of ni in terms of @xmath291 , with the constant factor in the quadratic convergence result given explicitly .
[ newquad ] for ni , ini_1 and ini_2 , we have @xmath330 for @xmath87 large enough .
for ni we have @xmath331 for @xmath87 large enough , that is , asymptotically , ni converges quadratically . since ni ,
ini_1 and ini_2 converge , for @xmath87 large enough we must have @xmath332 therefore , from ( [ eq : decomposition ] ) , the nonnegativity of @xmath10 and @xmath333 , for @xmath87 large enough we obtain therefore , we get @xmath334 since @xmath335 , we have @xmath336 as @xmath87 increases .
note that @xmath337 .
then from ( [ niconv ] ) we obtain @xmath338 for @xmath87 large enough , from which and ( [ evaerror ] ) it follows that ( [ evecerror ] ) holds . since @xmath339 , ( [ evecerror ] ) proves the ( asymptotic ) quadratic convergence of ni .
[ quadconvini ] for @xmath87 large enough , the inner tolerance @xmath340 in ini_2 satisfies @xmath341 and ini_2 converges superlinearly in the form of @xmath342 with the convergence order @xmath343 and @xmath27 a constant . by the condition of ini_2 and @xmath344 , we have @xmath345 note that ( [ evaerror ] ) holds for ini .
therefore , for @xmath87 large enough we have @xmath346 which is just ( [ innertol ] ) .
it is known from lemma [ coslower ] that @xmath228 for all @xmath87 .
for @xmath87 large enough , note that @xmath347 .
make the taylor expansion of the reciprocal of the denominator in ( [ eq : tan ] ) . then substituting ( [ evaerror ] ) into ( [ eq : tan ] ) and amplifying the term @xmath348 by some elementary manipulation we get @xmath349 with @xmath350 and @xmath351 certain positive constants .
since @xmath352 for @xmath87 large enough , the above inequality can be written as @xmath353 with @xmath27 a positive constant .
taking the equality sign in the above relation , by the theory of linear difference equation @xcite , for @xmath87 sufficiently large we obtain @xmath354 with @xmath355 and the final @xmath356 , which proves ( [ convorder ] ) .
we comment that if @xmath357 then @xmath358 in the above proof , in which case ini becomes ni and ( [ evecerror ] ) , the quadratic convergence of ni , is recovered .
( [ innertol ] ) indicates that the inner tolerance @xmath359 in ini_2 decreases like @xmath360 with increasing @xmath87 , so we may need to solve the inner linear systems more and more accurately as iterations proceed . as a compensation and gain , however ,
since ini_1 converges linearly , ini_2 may use fewer outer iterations to achieve the convergence than ini_1 .
a consequence is that it is hard and even impossible to compare the overall efficiency of ini_1 and ini_2 and draw a general definitive conclusion on which one of them is more efficient .
in this section , we consider how to compute the smallest eigenpair of an irreducible nonsingular @xmath0-matrix @xmath6 . in order to propose ini for this kind of problem ,
suppose that @xmath6 is expressed as @xmath361 , and let @xmath362 be the smallest eigenpair of it .
as have been proved previously , each of ini_1 and ini_2 generates a monotonically decreasing sequence @xmath200 that converges to @xmath14 with @xmath363 .
we denote @xmath364 .
it follows that @xmath365 and @xmath366 forms a monotonically increasing sequence that converges to @xmath2 . from algorithm alg : iiva , at iteration @xmath87 the approximate solution @xmath101 exactly solves the linear system@xmath367where @xmath368 .
we then update the next iterate as @xmath369 .
since@xmath370 ( [ eq : linearb ] ) is equivalent to @xmath371 where@xmath372 since @xmath373 is an irreducible nonsingular @xmath0-matrix , we have @xmath374 provided that @xmath375 .
thus , we get the relation @xmath376 therefore , algorithm [ alg : iiva ] can be adapted to computing the smallest eigenpair of the irreducible nonsingular @xmath0-matrix @xmath6 , which is described as algorithm [ alg : ivam ] , where , as in section [ sec : type1 ] , we define * ini_1 : the residual norm satisfies @xmath377 for some @xmath205 . * ini_2 : the residual vector satisfies @xmath206 with @xmath378 for @xmath379 and @xmath209 with some @xmath380
. we should point out that algorithm [ alg : ivam ] itself neither involves @xmath17 nor requires that @xmath6 be expressed as @xmath7 , which is purely for the algorithmic derivation . 1 .
given an initial guess @xmath381 with @xmath382 and @xmath76 , compute @xmath383 .
2 . * for * @xmath78 3 . solve @xmath384 approximately with the first or second inner tolerance strategy such that @xmath385 4 .
normalize the vector @xmath386 .
5 . compute @xmath387 .
* until * convergence : @xmath388 . due to the equivalence of algorithm [ alg : ivam ] and algorithm alg : iiva , the previous convergence results for the irreducible nonnegative matrix eigenvalue problem naturally hold for the irreducible nonsingular @xmath0-matrix eigenvalue problem under consideration .
we summarize the main results as follows .
let @xmath6 be an irreducible nonsingular @xmath0-matrix .
if @xmath389 and @xmath22 are generated by algorithm [ alg : ivam ] , then @xmath390 , the smallest eigenvalue of @xmath6 , monotonically from below as @xmath171 , and @xmath391 with @xmath392 for all @xmath393 .
furthermore , the convergence of ini_1 and ini_2 is globally linear with the asymptotic convergence factor bounded by @xmath23 and superlinear with the convergence order @xmath1 , respectively .
in this section we present numerical experiments to support our theoretical results on ini and ni , and illustrate the effectiveness of the proposed ini algorithms . in the meantime , we compare ini with the algorithms jdqr @xcite , jdrpcg @xcite , the implicitly restarted arnoldi method @xcite , i.e. , the matlab function , and the explicitly krylov schur method @xcite , whose matlab code is downloaded from @xcite , all of which are not positivity preserving for approximate eigenvectors .
we use jdqr for the nonnegative matrix and @xmath0-matrix eigenvalue problems , but we use jdrpcg only for symmetric nonsingular @xmath0-matrices since jdrpcg is designed to compute a few number of smallest eigenpairs of a symmetric matrix . we show that the ni and ini algorithms are always reliable to compute positive eigenvectors while the other algorithms generally fail to do so . actually , for the three ones of four problems tested , the converged eigenvectors obtained by the other algorithms were not positive . we also demonstrate that the ini algorithms are efficient , and they are competitive with and can be considerably efficient than the others .
all numerical tests were performed on an intel ( r ) core ( tm ) i@xmath394 cpu @xmath395@ @xmath396ghz with @xmath397 gb memory using matlab @xmath398 with the machine precision @xmath399 under the microsoft windows @xmath400 @xmath401-bit .
we present two examples to illustrate numerical behavior of ni , ini_1 and ini_2 for nonnegative matrices . at each outer iteration the approximate solution @xmath101 of ( [ eq : inexactsys ] ) satisfies@xmath402 by requiring the following inner tolerances : * for ni : @xmath403 ; * for ini_1 : @xmath404 with some @xmath405 ; * for ini_2 : @xmath406 for @xmath407 and @xmath408 with some @xmath409 .
we explain more on the inner tolerance used in ini_2 .
recall that we always have @xmath410 , which is reasonably small for @xmath134 very large .
therefore , starting with a general unit length vector @xmath411 , the sequence @xmath200 by ini_2 with @xmath412 may not satisfy @xmath413 , the condition of lemma [ monotone ] .
consequently , the global convergence of @xmath200 is not guaranteed .
however , we must have @xmath414 for the above - proposed inner tolerance for ini_2 , such that @xmath200 generated by ini_2 is globally convergent . furthermore ,
once @xmath73 has converged with certain accuracy , we will have @xmath415 for @xmath87 large enough .
after it , it is known from theorem [ quadconvini ] that ini_2 will asymptotically converge superlinearly . in implementations , it is necessary to impose some guards on inner tolerances so as to avoid them being too small in finite precision arithmetic , i.e. , below the level of @xmath416 .
for all examples , the stopping criteria for inner iterations are always taken as@xmath417and@xmath418 for each example , we test two values of @xmath419 to observe the effect of @xmath289 on the convergence of ini_1 . in the experiments , the stopping criterion for outer iterations is @xmath420 where we use the cheaply computable @xmath421 to estimate the 2-norm @xmath422 , which is more reasonable than the individual @xmath423 or @xmath424 with @xmath425 the infinity norm of a matrix . for ini and ni , since the coefficient matrices @xmath426 are always positive definite for all @xmath87 when @xmath10 is symmetric , we use the conjugate gradient method to solve inner linear systems . for @xmath10 unsymmetric , we use bicgstab as inner solver . in implementations
, we use the standard matlab functions and .
the outer iteration starts with the normalized vector @xmath427^t$ ] for ni , ini , , and .
the original codes of jdqr and jdrpcg use the absolute residual norms to decide the convergence , which is not robust for a general purpose . by setting the stopping criteria `` tol@xmath428 '' for outer iterations in them
, we will get the same stopping criteria as that for ni and ini .
we set the parameters `` sigma = lm '' and the inner solver `` options.lsolver=bicgstab '' in the unsymmetric case and `` options.lsolver=minres '' in the symmetric case , where the matlab function is the minimal residual method .
all the other options use defaults .
we do not use any preconditioning for inner linear systems . for and , we set the stopping criteria
opts.tol@xmath429 , and take the maximum and minimum subspace dimensions as @xmath430 and @xmath35 at each restart , respectively . for the computation of perron roots and vectors , we mention that and do not involve shift - invert , so we do not solve any inner linear systems .
we denote by @xmath431 the number of outer iterations to achieve the convergence and by @xmath432 the total number of inner iterations .
note that each outer iteration of these algorithms needs one matrix - vector product formed with @xmath10 , while each iteration of bicgstab uses two matrix - vector products formed with @xmath10 and @xmath433 , respectively , and each iteration of minres or cg uses one matrix - vector product with @xmath10 .
we denote by @xmath434 the total matrix - vector products formed with @xmath10 and possibly @xmath433 , which can fairly measure the overall efficiency of ni , ini , jdqr , and when the subspace dimensions used by jdqr , or are small at each restart . in view of the above , we have @xmath435 for @xmath10 unsymmetric and @xmath436 for @xmath10 symmetric for our test algorithms . for and we have @xmath437 since no inner linear system is solved .
recall that for unsymmetric problems jdqr , and may compute complex approximate perron vectors .
if it is the case , we are only concerned with the signs of components in its real part . to investigate the positivity of a converged perron vector
, we let the real part of its maximal component in magnitude be positive , no matter how real or complex it is . in the tables , positivityrecords whether the converged
perron vector preserves the strict positivity property .
if no , then the percentage in the brace indicates the proportion that the converged perron vector has the components with positive real part .
we also report the cpu time of each algorithm , which measures the overall efficiency too .
[ exp : google]from dimacs10 test set @xcite , we consider the unsymmetric nonnegative matrix .
the matrix data was released in 2002 by google as a part of google programming contest .
the is a directed graph with nodes representing web pages and directed edges representing hyperlinks between them .
this matrix is a binary matrix of order @xmath438 and has @xmath439 nonzero entries .
table [ table2 ] reports the results obtained by ni , ini_1 with @xmath440 and @xmath441 , ini_2 , jdqr , , and .
figures [ fig : google3][fig : google2 ] depict how the residual norms of outer iterations evolve versus the sum of inner iterations and versus the outer iterations for ni , ini_1 with two @xmath289 , ini_2 , and jdqr , respectively .
as we see , figure [ fig : google2 ] shows that the residual norms computed by the five algorithms decreased monotonically .
it also indicates that the ni and ini algorithms converged slowly and similarly in the beginning of outer iterations .
then they started converging fast .
we find that ini_1 and ini_2 achieved the same superlinear ( quadratic ) convergence and exhibited very similar convergence behavior to ni , and all of them used nine outer iterations to achieve the convergence .
these results show that our theory on both ini_1 and ini_2 can be conservative .
we observe from table [ table2 ] that the ini_1 and ini_2 improved the overall efficiency of ni considerably . actually , the cpu times used by ini_1 with @xmath442 and @xmath441 were @xmath443 and @xmath444 of those used by ni , respectively , and that used by ini_2 was @xmath443 of that used by ni . in terms of either @xmath434 or the cpu time , ini_1 with @xmath442 , ini_2 and jdqr were twice as fast as ni .
in addition , we find that ini_1 and ini_2 were competitive with , and they used very comparable cpu time .
table [ table2 ] also shows that used the least cpu time but more outer iterations than , and jdqr was as efficient as ini_1 and ini_2 in terms of @xmath434 and the cpu time .
the three algorithms jdqr , and computed the perron root reliably , but the converged perron vectors were not positive .
as table [ table2 ] indicates , only 41% , 42% and 21% of the components of the converged eigenvectors by jdqr , and were positive or had positive real part .
this implies that the too many components were not reliable and were hard to interpret .
we comment that the results by jdqr , and for a desired positive eigenvector are not unusual .
because some component(s ) of the perron vector must be very small , it is quite possible that the jdqr , and can not guarantee the strict positivity of approximate eigenvectors , and , particularly , those very small or tiny true positive components may change signs and become negative in the approximations , even though the approximations have already converged to the positive eigenvector @xmath3 in the conventional sense and attains its maximum accuracy , namely , the level of @xmath416 .
.the total outer and inner iterations in example exp : google [ cols="<,>,>,>,>,<",options="header " , ] .
the outer residual norms versus the outer iterations.,width=226 ] .
the outer residual norms versus the outer iterations.,width=226 ] we have also tested the previous four examples using the power method , which preserves the positivity of the approximate eigenvectors at each iteration and is linearly convergent for any positive starting vector . for examples [ exp : google][delaunay ] , we have found that the power method needed @xmath445 times of the cpu time of ini_1 because of the eigenvalue clustering . for @xmath0-matrices in examples [ exp:3dmesh ] and [ exp : nicolo ] , we applied the power method to the nonnegative matrices @xmath10 , which were generated by @xmath446 with @xmath17 chosen as before , for finding @xmath14 and the perron vectors .
the power method cost more than ten times of the cpu time of ini_1 .
so ini was much more efficient than the power method for these four examples .
we omit the details on numerical results obtained by the power method .
for the efficient computation of the smallest eigenpair of a large irreducible nonsingular @xmath0-matrix , we have proposed a positivity preserving inexact noda iteration method with two practical inner tolerance strategies provided for solving the linear systems involved .
we have analyzed the convergence of the method in detail , and have established a number of global linear and superlinear convergence results , with the linear convergence factor and the superlinear convergence order derived explicitly .
precisely , we have proved that ini converges at least linearly with the asymptotic convergence factor bounded by @xmath23 for @xmath447 and superlinearly with the convergence order @xmath1 for the decreasing @xmath448 , respectively .
we have also revisited the convergence of ni and proved its quadratic convergence in a different form from @xcite .
the results on ini clearly show how inner tolerance affects the convergence of outer iterations .
numerically , we have illustrated that the proposed ini algorithms are practical for large nonnegative matrix and @xmath0-matrix eigenvalue problems , and they can reduce the total computational cost of ni substantially . in the meantime , regarding the positivity preservation , the numerical experiments have shown that the ini algorithms are superior to the jacobi davidson method , the implicitly restarted arnoldi method and the krylov schur method .
they always preserve the positivity of approximate eigenvectors , while the jacobi davidson method , the implicitly restarted arnoldi method and the krylov
schur method often fail to do so .
moreover , regarding he overall efficiency , ini algorithms are competitive with and can be considerably higher than the other three methods for some practical problems .
aim@shape : advanced and innovative models and tools for the development of semantic - based systems for handling , acquiring , and processing knowledge embedded in multidimensional digital objects .
available at ` http://shapes.aim-at-shape.net ` x. d. gu , t .-
huang , w .- q .
huang , s .- s .
lin , w .- w .
lin and s .- t .
yau , high performance computing for spherical conformal and riemann mappings , technical report , department of applied mathematics , national chiao tung university , taiwan ( 2013 ) a. m. ostrowski , on the convergence of the rayleigh quotient iteration for the computation of the characteristic roots and vectors . v. ( usual rayleigh quotient for non - hermitian matrices and linear elementary divisors ) , arch
. rational mech .
, 3 , pp .
472481 ( 1959 ) p. shivakumar , j. williams , q. ye and c. marinov , on two - sided bounds related to weakly diagonally dominant m - matrices with applications to digital circuit dynamics , siam j. matrix anal .
appl . , 17 , pp . 298312 ( 1996 ) m. q. wei , m. y. pang and c. l. fan , survey on planar parameterization of triangular meshes , proceedings of the 2010 international conference on measuring technology and mechatronics automation , pp . 702705 ( 2010 ) f. xue , d.b .
szyld , efficient preconditioned inner solves for inexact rayleigh quotient iteration and their connections to the single - vector jacobi davidson method , siam j. matrix anal .
32 , pp . 9931018 ( 2011 )
|
in this paper , based on the noda iteration , we present inexact noda iterations ( ini ) , to find the smallest eigenvalue and the associated positive eigenvector of a large irreducible nonsingular @xmath0-matrix .
the positivity of approximations is critical in applications , and if the approximations lose the positivity then they may be meaningless and could not be interpreted .
we propose two different inner tolerance strategies for solving the inner linear systems involved , and prove that the convergence of resulting ini algorithms is globally linear and superlinear with the convergence order @xmath1 , respectively .
the proposed ini algorithms are structure preserving and maintains the positivity of approximate eigenvectors .
we also revisit the exact noda iteration and establish a new quadratic convergence result .
all the above is first done for the problem of computing the perron root and the positive perron vector of an irreducible nonnegative matrix and is then adapted to computing the smallest eigenpair of the irreducible nonsingular @xmath0-matrix .
numerical examples illustrate that the proposed ini algorithms are practical , and they always preserve the positivity of approximate eigenvectors .
we compare them with the jacobi davidson method , the implicitly restarted arnoldi method and the explicitly restarted krylov
schur method , all of which can not guarantee the positivity of approximate eigenvectors , and illustrate that the overall efficiency of the ini algorithms is competitive with and can be considerably higher than the latter three methods .
example.eps gsave newpath 20 20 moveto 20 220 lineto 220 220 lineto 220 20 lineto closepath 2 setlinewidth gsave .4 setgray fill grestore stroke grestore
|
random walks on graphs serve as the basis of algorithms for solving a number of problems , including 2-sat , graph connectivity , and finding satsifying assignments for boolean functions . quantum algorithms
have shown promise in solving some problems faster than is possible using classical algorithms .
quantum walks represent an attempt to `` quantize '' classical random walk algorithms and thereby increase their speed .
quantum algorithms must be run on a quantum computer . in this kind of machine
, information is represented not in terms of bits , but in terms of quantum bits , or qubits .
a bit is either @xmath1 or @xmath2 and can be represented by current on or off , or by two different voltages .
a qubit is a two - level quantum system , and possibilites include the spin states of an electron or the polarization states of a photon .
one of the states , e.g. spin up , corresponds to @xmath1 and the other , e.g. spin down , corresponds to @xmath2 .
the principle difference between a bit and a qubit , is that the bit is either @xmath1 or @xmath2 , while the qubit can be in a superposition of its @xmath1 state and its @xmath2 state .
computers based on bits represent and process information according to the rules of classical physics , while computers based on qubits represent and process information according to the rules of quantum physics .
there has been some progress in using quantum walks to construct quantum algorithms .
it has recently been shown that it is possible to use a quantum walk to perform a search on the hypercube faster than can be done classically @xcite . in this problem
the number of steps drops from @xmath3 , which is the number of vertices , in the classical case to @xmath4 in the quantum case .
a much more dramatic improvement has recently been obtained by childs , _ et al .
they constructed an oracle problem that can be solved by a quantum algorithm based on a quantum walk exponentially faster than is possible with any classical algorithm .
quantum walks were introduced by y. aharanov , l. davidovich and n. zagury @xcite .
they considered a spin-@xmath5 particle that is shifted to the left or right depending on its spin state . since then , a number of different kinds of quantum walks have been developed .
the first , the coined walk , is a discrete time walk that makes use of an auxiliary quantum system , the coin .
this is necessary to make the operator that advances the walk one step unitary .
this walk was originally proposed by watrous and is closlely related to the original walk discussed by aharanov @xcite .
a second type of walk is the continuous time walk , due to childs , _
et al_. @xcite . here
the hamiltonian that is used to construct the time development transformation is taken to be proportional to the adjancency matrix of the graph .
a third type of walk , based on thinking about the graph as an interferometer with optical multiports as the nodes was recenty proposed by us @xcite . in this case the walk takes place on the edges of the graph , rather than the vertices , and each edge has two states , one corresponding to traversing the edge in one direction and the other to traversing the edge in the opposite direction .
it should be noted that while it is very simple to construct a quantum walk on an arbitrary graph using the second and third methods , it is much less obvious how to do so for the coined walk
. let us briefly describe the coined quantum walk . in trying to formulate a quantum walk on a graph
, the most natural thing to do is to let a set of orthonormal basis states correspond to the vertices of the graph .
if a particle is in the state @xmath6 , that corresponds to its being located on vertex @xmath0 .
let the hilbert space spanned by these states be @xmath7 .
trying to define a unitary evolution using this scheme soon leads to serious problems , as was first noted by meyer @xcite .
watrous solved this problem by enlarging the hilbert space in which the quantum walk takes place .
how this scheme works is most easily seen by considering the quantum walk on a line .
the vertices are labelled by integers , and , in addition , there is a quantum coin , which has two states , @xmath8 and @xmath9 , corresponding to left and right , respectively , which span a two - dimensional hilbert space , @xmath10 .
a basis for the hilbert space describing the entire system , @xmath11 , is given by the states @xmath12 , where @xmath0 is an integer , and @xmath13 is either @xmath14 or @xmath15 .
a step in this walk consists of applying the hadamard operator , @xmath16 , to the coin , @xmath17 and then the operator @xmath18 where @xmath19 is the shift operator , whose action is given by @xmath20 note that the evolution of the quantum state of the walk is deterministic , that is , if the initial state is @xmath21 , then the state after @xmath22 steps is just @xmath23^{m}|\psi_{in}\rangle$ ] . probabilities enter the picture when we try to determine where the particle is by making measurements .
coined quantum walks have been studied on both the line by nayak and vishwananth @xcite and on the cycle by d. aharanov _ et al_. @xcite .
et al_. also studied a number of properties of quantum walks on general graphs .
absorbing times and probabilities for walks on the line have been examined by several authors @xcite .
going beyond one dimension , properties of quantum walks on two and three dimensional lattices @xcite and on the hypercube @xcite have been explored .
the effects of decoherence on quantum walks have also been studied .
et al_. showed how increasing amounts of decoherence turn a quantum walk into a classical random walk @xcite .
kendon and tregenna found that small amounts of decoherence can speed up the convergence of the time - averaged probability distribution of a quantum walk @xcite .
many aspects of quantum walks are covered in the recent and excellent review by kempe @xcite .
several proposals have been made for the physical realization of quantum walks @xcite-@xcite .
the ones closest in spirit to the walks studied here are the realizations that employ optical methods , either linear optical elements @xcite or cavities @xcite .
the last two references show that an experimental quantum walk has , in fact , been carried out , though it was not intrepreted as such at the time @xcite . here
we wish to describe quantm walks that take place on the edges of a graph , and to study the connection between these walks and scattering theory . in order to show how scattering theory enters the picture , begin with a graph ,
choose two vertices , and attach two semi - infinite lines , which we shall call tails , to it , one to each of the selected vertices .
each of the half - lines is made up of vertices and edges .
thus we can start the walk on one of the tails , have it progress into the original graph , and emerge onto the opposite tail .
this type of arrangement allows us to define a scattering matrix , or s matrix , for the original graph , with the amplitude to get from one tail to the other being called the transmission amplitude .
as we shall see properties of quantum walks starting on one tail and ending on the other can be expressed in terms of the transmission amplitude of the graph .
it should be noted that scattering approaches have also proven useful in a related area , quantum graphs , which can be used to study electron transport in large molecules @xcite .
the first use of scattering theory to analyze the behavior of quantum walks was done by farhi and gutmann @xcite .
they studied continuous - time walks on trees , and they added tails to the trees to place the walks into a scattering - theoretic framework .
they were able to place bounds on the time for a walk , starting at the root of the tree , to penetrate to one of the leaves .
we begin by defining our quantum walk , and then illustrate it with an example .
we then explore the relation between properties of the quantum walk and the transmission amplitude of the graph .
these are then applied to an example .
we define a time - reversal transformation for quantum walks and use it to explore when the transmission amplitude to go through the graph in one direction is the same as the transmission amplitude to go through it in the other direction . finally , we finish with a detailed study of the properties of the transmission and reflection amplitudes , in particular showing that they are analytic in on the unit circle and in its interior .
this allows us to express the probability of starting on one tail and finishing on the other in @xmath0 steps and also the hitting time for a walk from one tail to the other very simply in terms of the transmission amplitude of the graph .
the type of quantum walk that will be used throughout this paper was originally presented in @xcite . we imagine a particle on an edge of a graph ; it is this particle that will make the walk .
each edge has two states , one going in one direction , the other going in the other direction , and we denote the set of oriented edges ( an edge and its direction of traversal is an oriented edge ) of the graph by @xmath24 .
that is , if our edge is between the vertices @xmath25 and @xmath26 , which we shall denote as @xmath27 , it has two orthogonal states , @xmath28 , corresponding to the particle being on @xmath27 and going from @xmath25 to @xmath26 , and @xmath29 , corresponding to the particle being on @xmath27 and going from @xmath26 to @xmath25 .
the collection of all of these edge states is a basis for a hilbert space , and the states of the particle making the walk lie in this space , which is @xmath30 . now that we have our state space , we need a unitary operator that advances the walk one step .
let us first illustrate how this works for a walk on the line .
we shall label the vertices by the integers . in this case , the states of the system are @xmath31 , where @xmath32 .
the vertices can be thought of as scattering centers .
consider what happens when a particle , moving in one dimension , hits a scattering center .
it has a certain amplitude to continue in the direction it was going , i.e. to be transmitted , and an amplitude to be change its direction , i.e. to be reflected .
the scatterer has two input states , the particle can enter from either the right or the left , and two output states , the particle can leave heading either right or left .
the scattering center defines a unitary transformation between the input and output states .
we now need to translate this into transition rules for our quantum walk .
suppose we are in the state @xmath33 .
if the particle is transmitted it will be in the state @xmath34 , and if reflected in the state @xmath35 .
let the transmission amplitude be @xmath36 , and the reflection amplitude be @xmath37 .
we then have the transition rule @xmath38 where unitarity implies that @xmath39 .
the other possibility is that the particle is incident on vertex @xmath40 from the right , that is it is in the state @xmath41 .
if it is transmitted it is in the state @xmath35 , and if it is reflected , it is in the state @xmath34 .
unitarity of the scattering transformation then gives us that @xmath42
these rules specify our walk . the case @xmath43 and @xmath44 corresponds to free particle propagation ; a particle in the state @xmath34 simply moves one step to the right with each time step in the walk .
if @xmath45 , then there is some amplitude to move both to the right and to the left . as an example , let us consider the case when @xmath46 , and the particle starts in the state @xmath47 .
the probability distribution for finding the particle on a given edge after @xmath48 steps is shown in figure 1 .
first , note that this distribution look nothing like the distribution that would result from a classical random walk on a line , which would be a gaussian centered about the initial position of the particle .
in addition , we see that there is an interval about the origin in which it is very likely to find the particle .
an asymptotic analysis of the walk shows that this region lies between @xmath49 and @xmath50 , where @xmath0 is the number of steps in the walk and @xmath36 is the transmission amplitude of the vertices @xcite .
that is , the region in which the the particle is likely to be found grows as @xmath0 instead of @xmath51 as would be the case in a standard random walk .
[ fig1 ] thinking about the particle making the walk and the scattering centers as beam splitters , we see that this type of quantum walk is analogous to the propagation of light through an interferometer .
this suggests that we can add a new element to quantum walks that has no analogue in classical random walks .
interferometers are made up of multiports ( generalized beam splitters ) @xcite and phase shifters ; a phase shifter imparts a constant phase to a photon that passes through it .
suppose we were to put a phase shifter that imparts a phase shift of @xmath52 just before the @xmath53 vertex .
the transition rules for the states adjacent to this vertex are modified , while the rules for all other states are unaffected .
in particular , we now have @xmath54 insertion of a phase shifter into an edge can change the properties of a quantum walk , because it changes how different paths interfere .
so far we have only considered vertices at which two edges meet , but if we are to construct graphs more complicated than lines , we need to see how a vertex with more that two edges emanating from it behaves .
we shall begin by looking at an example .
if a vertex treats all edges entering it in an equivalent fashion , then we have a particulary simple situation , because the edges of the graph do not have to be labelled .
one vertex of this type is very closely related to the quantum coin used in the walk on the hypercube @xcite .
let the vertex at which all of the edges meet be labelled by @xmath55 , and the opposite ends of the edges be labelled by the numbers @xmath2 through @xmath0 .
for any input state , @xmath56 , where @xmath57 is an integer between @xmath2 and @xmath0 , the transition rule is that the amplitude to go the output state @xmath58 is @xmath37 , and the amplitude to go to any other output state is @xmath36 .
that is , the amplitude to be reflected is @xmath37 , and the amplitude to be transmitted through any of the other edges is @xmath36 .
unitarity places two conditions on these amplitudes @xmath59 we shall call such vertices equal - transmission vertices . as an example , for the case @xmath60 , possible values of @xmath37 and @xmath36 are @xmath61 and @xmath62 . in order to construct a walk for a general graph ,
one chooses a unitary operator for each vertex , i.e. one that maps the states coming into a vertex to states leaving the same vertex .
more precisely , let @xmath63 be the set of oriented edges starting at the vertex @xmath25 and @xmath64 be the linear span of @xmath63 .
similarly , let @xmath65 be the set of oriented edges ending at the vertex @xmath25 , and @xmath66 be the linear span of @xmath65 .
we have that @xmath67 and @xmath68 if @xmath69 , which implies that @xmath70 and @xmath71 where the unions and direct sums are over all vertices of the graph .
the local unitary operator , @xmath72 maps @xmath66 to @xmath64 , and describes the scattering at vertex @xmath25 .
one step of the walk consists of the combined effect of all of these operations ; the overall unitary operator , @xmath73 , that advances the walk one step is constructed from the local operators for each vertex .
explicitly , the edge state @xmath28 , which is the state for the particle going from vertex @xmath25 to vertex @xmath26 , will go to the state @xmath74 after one step , where @xmath75 is the operator corresponding to vertex @xmath26 .
this prescription guarantees that the overall operation is unitary , in particular , @xmath73 acting on any other edge state @xmath76 will give a state orthogonal to @xmath77 .
if @xmath78 , then @xmath28 and @xmath76 will be mapped into @xmath79 , but the unitarity of @xmath75 ensures that @xmath77 and @xmath80 are orthogonal .
if @xmath81 , then @xmath73 maps @xmath28 and @xmath76 onto different sets of states , @xmath79 and @xmath82 , respectively , and the results are then orthogonal . therefore , as the edge states make up an orthonormal basis of the hilbert space in which the walk occurs , and @xmath73 maps this basis to another orthonormal basis , it is unitary .
let us put all of this together in a very simple example .
consider the graph shown in figure 2 , where each of the vertices where two edges meet have @xmath43 and @xmath44 , while the three - edge vertices are of the type discussed in the previous section , with @xmath61 and @xmath62 .
there is a phase shifter with an adjustable phase , @xmath52 , in one of the edges .
the graph goes to negative infinity on the left and plus infinity on the right .
note that it is very simple to construct the unitary operator that advances a quantum walk on this graph by one step ; one simply combines the actions of the operators corresponding to each vertex .
[ fig2 ] we can find the unnormalized eigenstates for this graph , and one set of them can be described as having an incoming wave from the left , an outgoing transmitted wave going to the right , and a reflected wave going to the left .
a second set will have an incoming wave from the right , an outgoing trasmitted wave to the left , and a refelected wave to the right .
the mathematical status of these objects will be explored in sections 7 and 8 . finally , there may be bound states , i.e. eigenstates that are localized in the region between the two vertices with three edges .
now , let us denote the left three - edge vertex by @xmath1 and the right one by @xmath83 , and number the vertices on the lines correspondingly , from @xmath83 to plus infinity to the right and from @xmath1 to minus infinity to the left .
the eigenstates with a wave incident from the left take the form @xmath84 where @xmath85 is the part of the eigenfunction between vertices @xmath1 and @xmath83 , and @xmath86 is the eigenvalue of the operator @xmath73 that advances the walk one step .
the first term can be thought of as the incoming wave ; it is confined to the region between negative infinity and @xmath1 , and consists of states in which the particle is moving to the right .
the term proportional to @xmath87 is the reflected wave .
it is also confined to the region between negative infinity and @xmath1 , but consists of states in which the particle is moving to the left . finally ,
the term proportional to @xmath88 is the transmitted , being confined to the region from @xmath83 to infinity , and consisting of states with the particle moving to the right .
the coefficient @xmath87 can be interpretted as the amplitude for the incoming wave to be reflected , and @xmath88 as the amplitude to be transmitted . denoting the upper vertex between @xmath1 and @xmath83 as @xmath89 and the lower one as @xmath90
, we can express @xmath85 as @xmath91 the solution of the equation @xmath92 is given in the appendix .
one can define a transmission coefficient for this graph , which is just the ratio of the intensities of the outgoing transmitted and the incoming waves , @xmath93 one finds that for @xmath94 , t is nonzero , because the two paths from @xmath1 to @xmath83 interfere constructively , while when @xmath95 they interfere destructively , which results in @xmath96 .
therefore , the behavior of the walk strongly depends on the value of the phase shift .
suppose we want to find out whether the particle making the walk is on the edge between vertices @xmath40 and @xmath97 . the relevant projection operator is @xmath98 if we obtain @xmath2 the particle is on that edge , and if we obtain @xmath1 , it is not .
what we wish to find out is , if the particle is initially in the state @xmath99 , the probability that the particle is not on the edge for the first @xmath0 steps of the walk , but is on the @xmath100st .
we measure @xmath101 after every step , and we have to take the effect of these meaurements into account in describing the dynamics of the wave function .
let us first see what happens after one step . let @xmath73 be the unitary operator that advances the walk one step .
the probability that the particle is not on the edge between @xmath40 and @xmath97 after one step , @xmath102 , is @xmath103 and the quantum state , assuming the particle was not found on this edge , is @xmath104 now let us go one more step to see the pattern .
the probability that the particle is not on the edge between @xmath40 and @xmath97 after either the first or second steps , @xmath105 , is @xmath106 and the quantum state , assuming the particle was not found on the edge after either step , is @xmath107 in general , if @xmath108 is the probability of not finding the particle on the edge between @xmath40 and @xmath97 after each of @xmath0 steps , then @xmath109 where @xmath110 is the quantum state after @xmath111 steps and @xmath111 measurements indicating that the particle was not on the designated edge .
we also have that @xmath112 let us now proceed by induction .
we shall assume that @xmath113^{n}\psi ( 0)\|^{2 } \nonumber \\
{ \| [ ( i - p_{j})u]^{n}\psi ( 0)\| } , \end{aligned}\ ] ] which clearly holds for @xmath114
. substitution of the second of these equations into eq .
( [ staten ] ) yields @xmath115 so that part of our induction hypothesis is verfied . from eq .
( [ probn ] ) we find that @xmath116^{n}\psi ( 0)\|^{2 } \| ( i - p_{j})u \psi ( n)\|^{2 } \nonumber \\ & = & \| [ ( i - p_{j})u]^{n+1}\psi ( 0)\|^{2 } , \end{aligned}\ ] ] which proves the remaining part .
this result is easily generalized to the case of making measurements to determine whether the particle is on a set of edges instead of on a single edge .
a similar result for the coined quantum walk was derived by bach , _ et al_. @xcite .
the hitting time for a random walk on a graph is the expected number of steps in a walk that starts at one specified vertex and stops on first reaching a second specified vertex . in order to calculate this
we need the probability that a walk starting on the first vertex does not reach the second for its first @xmath111 steps , but does reach it on the @xmath0th step .
the corresponding quantity for a quantum walk , which we shall call @xmath117 , is the probability that a walk starting in the state @xmath99 is not measured to be on the edge between @xmath40 and @xmath97 for the first @xmath111 steps , but is measured to be on that edge at the @xmath0th step .
this probability is given by @xmath118^{n-1}\psi ( 0)\|^{2 } .\end{aligned}\ ] ] the particle making this walk may never be found to be on the edge between @xmath40 and @xmath97 at all , and the probability that it is , is given by @xmath119 we can define a conditional hitting time for this walk , @xmath120 , as @xmath121 this quantity provides an an indication of how many steps a walk that does reach the edge between @xmath40 and @xmath97 will require .
let us now consider the situation depicted in figure 3 .
we start the walk on the edge @xmath122 moving to the right , and we are interested in how long it takes to get to the edge @xmath34 , where @xmath123 is a general graph with a finite number of edges and vertices .
the vertices outside of the graph , @xmath123 , for example , @xmath124 and @xmath97 , have a transmission amplitude , @xmath43 .
we shall denote the entire graph , @xmath123 plus the tails , as @xmath125 . the case considered in
the previous section is a special case of this situation .
when calculating the probability , @xmath117 , that the walk arrives at the edge @xmath34 for the first time after @xmath0 steps , we need to evaluate ( see eq .
( [ qn ] ) ) @xmath126^{n-1}=p_{j}u^{n}-\sum_{k=1}^{n-1}p_{j}u^{(n - k ) }
p_{j}u^{k } + \ldots\ ] ] we first note that once the particle leaves the graph , @xmath123 , it moves one unit to the right with each step .
that means that terms of the form @xmath127 , for @xmath128 , are zero , because the first @xmath101 puts the particle in the state @xmath34 , and the operator @xmath129 moves it @xmath22 steps to the right , so that the particle in now in the state @xmath130 .
the projection @xmath101 acting on this state gives zero .
therefore , only the first term on the right - hand side of eq .
( [ out ] ) gives a nonzero contribution .
that means that in this case , we have @xmath131 [ fig3 ] we now want to relate the probability @xmath117 to the transmission coefficient of the graph , which we shall define shortly
. the eigenstates of the system can be grouped into three sets , those with the particle coming in from the left , those with the particle coming in from the right , and bound states .
the eigenstates of this system corresponding to the particle coming in from the left are @xmath132 is the transmission coefficient , and @xmath133 is the part of the eigenfunction inside the graph @xmath123 .
this state is , in fact , not in @xmath30 , and its mathematical significance will be clarified in a later section . in appendix
b we show that latexmath:[\[\label{reftrans } for the initial state of our walk we will choose @xmath122 , and we want to express this state in terms of the eigenstates in eq .
( [ geneigen ] ) . to do
so we integrate @xmath135 from @xmath1 to @xmath136 , finding @xmath137.\end{aligned}\ ] ] by setting @xmath138 the integrals on the right - hand side of the above equation can be turned into integrals along the unit circle , which we shall denote by @xmath139 , in the complex plane .
these integrals are of the form @xmath140 where @xmath141 .
as we shall show in section 7 , @xmath142 , @xmath143 , and @xmath144 are analytic functions of @xmath145 on the unit circle and in its interior , and they all vanish when @xmath146 .
this implies that all of the above integrals are zero .
therefore , we can conclude that @xmath147 now we have that @xmath148 and @xmath149 therefore , we find @xmath150 which relates a walk probability , the probability to land on the edge between @xmath40 and @xmath97 for the first time after @xmath0 steps , to the s matrix of the graph @xmath123 .
we can rewrite the integral appearing in the above expression for @xmath117 as a contour integral over the unit circle , @xmath139 by setting @xmath151 .
this gives us @xmath152 because , as we shall see , @xmath143 is analytic in a region containing the unit circle and its interior , the only contribution to the integral in eq .
( [ integral ] ) comes from the pole at the origin that is due to the @xmath153 factor .
this fact makes it straightforward to evaluate the above integral , by just finding the residue of the pole at the origin , giving us @xmath154 this is just the @xmath155 coefficient in the taylor expansion of @xmath143 about the origin , i.e. if @xmath156 then we have that @xmath157 .
the conditional hitting time , defined in eq .
( [ hit ] ) , is then given by @xmath158 we now note that @xmath159 from which it follows , setting @xmath160 and integrating , that @xmath161 similarly , we have that @xmath162 the integrals in the previous two equations can also be expressed as contour integrals if we define the `` reflected '' transmission amplitude , @xmath163 to be @xmath164 .
we then have that @xmath165 thus , we see that both the probability of first hitting the edge between @xmath40 and @xmath97 in @xmath0 steps and the hitting time to reach this edge are related to the transmission amplitude of the graph .
let us now apply this formula to a walk on the graph discussed in section 3 .
the transmission amplitude for this graph is @xmath166 and @xmath167 from this expression we see explicitly that @xmath143 is analytic inside and on the unit circle . considering the case @xmath94 we find the simple expression @xmath168 inserting this result into the expressions in the previous section , it is straightforward to perform the integrations , and we find that @xmath169 for small @xmath0 , it is relatively straightforward to verify the expression for @xmath117 by adding up the amplitudes for the possible paths from @xmath122 to @xmath170 of length @xmath0 .
another issue that deserves mention is the existence of bound states . in standard quantum scattering theory , if a potential has bound states , these are orthgonal to all scattering states .
what we mean by a bound state of a graph , is an eigenstate of @xmath73 whose support is confined to @xmath123 .
if our graph has bound states , then these states could not be accessed by a particle coming in from one of the tails , e.g. from the left .
this follows from the fact that the initial state is orthogonal to any bound state ( a state localized on one of the tails is automatically orthogonal to one that is localized in @xmath123 ) , and if @xmath171 is the initial state , and @xmath172 is the localized eigenstate with eigenvalue @xmath173 , we have @xmath174 this limits the walks that particles with initial states of this type can perform , because , as the above equation shows , the state of the walk at any time must be orthogonal to all of the bound states . in order to find the bound states in our example we consider a state @xmath175 that is located on the four edges connecting the vertices @xmath1 , @xmath89 , @xmath90 , and @xmath83 , i.e. @xmath91 we then solve the equation @xmath92 .
this equation only has a solution when @xmath94 , and in that case the eigenvalues are @xmath176 , where @xmath177 . the eigenstates in this case
are given by @xmath178 a simple basis for the space spanned by these eigenstates is latexmath:[\ ] ] where the sums are over all edges in @xmath123 .
we are first interested in determining whether @xmath143 is analytic inside the unit circle
. this will be the case if eq .
( [ ginhom ] ) can be solved for all @xmath244 , which will be true if the matrix @xmath245 is invertible for @xmath244 .
this follows from the result that if an operator , @xmath246 , is analytic in some region of the complex plane , so is its inverse , @xmath247 provided it exists @xcite . because @xmath248 we have that @xmath249 , which implies that if @xmath244 , then @xmath245 is invertible .
we now have that @xmath143 and @xmath142 are analytic inside the unit circle , but what about on the circle itself and a bit beyond ? to show that they are , we need to refine our arguments .
1 . @xmath250 if and only if @xmath251 is an @xmath252 eigenstate of @xmath73 with support in the edges of @xmath123 .
we call these bound states .
2 . if @xmath253 , then @xmath254 , where @xmath251 is any bound state .
1 . @xmath255 + so @xmath256 and the support of @xmath251 lies in the edges of @xmath123 .
therefore , @xmath257 , and @xmath251 is the desired eigenstate .
2 . this follows from @xmath258 if @xmath86 corresponds to a bound state , then the operator @xmath259 is not invertible . at the moment this leaves us without a method of defining @xmath143 and @xmath142 for values of @xmath145 corresponding to bound states .
we would like to show , however , that for these values of @xmath145 , scattering solutions , that is solutions to eq .
( [ ginhom ] ) , also exist .
these solutions can then be used to define @xmath143 and @xmath142 .
let @xmath260 be the eigenstates of @xmath73 , where @xmath261 and @xmath262 .
let @xmath263 be the span of @xmath264 .
@xmath263 is an invariant subspace of both @xmath73 and @xmath265 .
let @xmath266 as a subspace of @xmath206 , and let @xmath267 be the orthogonal projection of @xmath206 onto @xmath268 , and @xmath269 .
we wish to solve the equation @xmath270 on @xmath268 . because of the construction of @xmath268
, we are guaranteed that there exists an @xmath271 such that @xmath272 is invertible on @xmath268 for @xmath273 .
this allows us to define @xmath274 and @xmath275 we shall summarize our results in a theorem . 1 . the complex - valued functions @xmath142 and @xmath143 , and the function @xmath144 , which takes values in @xmath268 , are analytic functions on the disc @xmath276 .
2 . for each @xmath277 , the punctured disc
, @xmath278 is an element of @xmath279 , and @xmath278 is an analytic function of @xmath145 on the domain @xmath280 with values in @xmath279 .
@xmath278 is the solution to the generalized eigenvalue equation @xmath281 .
let us now make two brief observations .
first , note that the functions @xmath282 and @xmath238 are just the restrictions of @xmath143 and @xmath142 to the unit circle .
second , we see that the above theorem , along with the expressions for @xmath143 and @xmath142 , provides the necessary conditions for all of the integrals in eq .
( [ int0 ] ) to vanish .
the formal eigenvectors @xmath283 , which we just constructed , are elements of @xmath279 , and hence , for any @xmath284 , the inner product @xmath285 is well - defined , because it is absolutely summable .
however , if @xmath286 is merely in @xmath30 , @xmath285 is not necessarily summable . due to the specific form of @xmath287 , this sum represents a fourier series with square - summable coefficients , and
hence it is a function in @xmath288 , where @xmath289 is the unit circle with the usual measure @xmath290 .
we shall now exploit these two interpretations of our formulas .
recall that @xmath263 is the space of bound states , i.e. @xmath252 eigenfunctions of @xmath73 .
@xmath263 is a finite - dimensional , @xmath73-invariant subspace of @xmath207 .
each element of @xmath263 has its support contained among the edges of @xmath123 .
the function ( vector ) @xmath291 constructed in the last section is seen to be the unique element of @xmath279 such that , 1 .
@xmath292 , 2 .
@xmath293 , 3 .
@xmath294 for any @xmath295 . the analytic extension of @xmath287 to @xmath296
is given by @xmath278 ( see eq . ( [ psi(z ) ] ) ) , and a direct application of the cauchy integral theorem to @xmath278 implies that @xmath297 if @xmath298 , then @xmath285 is a continuous function on the circle , and @xmath299 in particular , if @xmath300 , we have @xmath301 from this we have that @xmath302 hence , for each @xmath286 in the linear span of the states @xmath303 , where @xmath57 is an integer , and for any trigometric polynomial @xmath304 , we have the @xmath305 formulas @xmath306 eq .
( [ c ] ) is an @xmath305 formula , however , eq . ( [ d ] ) is more general .
it holds for any @xmath286 that is a finite linear combination of the @xmath303 , for @xmath57 an integer .
however , @xmath307 is an @xmath252 function on the circle for each @xmath308 .
let @xmath309 be the closure of the linear span of the @xmath303 in @xmath201 .
@xmath309 is a closed @xmath73-invariant subspace of @xmath207 , and we see that eq .
( [ d ] ) holds for @xmath310 and for any complex - valued continuous function on @xmath289 .
thus , by the riesz - markov theorem there exists a unique measure @xmath311 , a borel measure on the circle , called the spectral measure associated to @xmath286 , such that @xmath312 and hence , @xmath313 if we define @xmath314 by @xmath315 then eq . ( [ d ] ) implies that @xmath316 , so that @xmath317 is an isometry from @xmath309 to @xmath288 , and @xmath318 is a closed subspace of @xmath288 .
finally , we have @xmath319 for @xmath320 which implies that @xmath318 is an invariant subspace of the multiplication operator @xmath321 .
therefore , we have that @xmath322 , and hence @xmath317 is a unitary operator of @xmath309 onto @xmath288
. we can summarize our results to date in the following theorem .
let @xmath309 be the closed , linear , u - invariant subspace of @xmath201 generated by the states @xmath303 , for @xmath57 an integer .
let @xmath287 be the generalized eigenfunctions constructed in the previous section , then for each @xmath310 , @xmath315 defines a unitary operator from @xmath309 onto @xmath288 , such that for each @xmath323 , @xmath324 hence , we have constructed part of the spectral decomposition of @xmath73 .
let @xmath325 be the closed , @xmath73-invariant subspace of @xmath207 generated by the vectors @xmath326 , for @xmath57 an integer .
in the same way as was done in the previous section , we can construct generalized eigenfunctions @xmath327 which satisfy the results of the last theorem of the previous section .
if we follow the arguments above , we arrive at the following theorem .
let @xmath325 and @xmath328 be as above .
then for each @xmath329 , @xmath330 defines a unitary operator from @xmath325 onto @xmath288 , such that for each @xmath323 , @xmath331 it has already been shown and is easy to see that @xmath309 , @xmath325 , and @xmath263 are mutually orthogonal , @xmath73-invariant subspaces of @xmath201 .
the final result of this section is that these subspaces exhuast @xmath207 , and , therefore , this completes the spectral decomposition of @xmath73 on @xmath207 .
@xmath332 .
there are many ways to prove this .
the following is in the spirit of our original argument .
if @xmath333 , then we can find a generalized eigenfunction @xmath287 , such that @xmath334 where @xmath287 has no overlap with @xmath122 , @xmath41 , or @xmath263 .
then a direct calculation shows that @xmath287 viewed as a function defined as on the edges , @xmath24 , is identically zero .
we have explored a number of properties of quantum walks on graphs .
the main result is that there is a connection between the s matrix of a graph and the properties of a quantum walk on that graph . in particular , the probability that a quantum walk traverses a graph in @xmath0 steps and the conditional hitting time to make a walk traversing a graph can be expressed in terms of the transmission amplitude of that graph .
this required studying the analyticity properties of the reflection and transmission coefficients of the graph .
in addition , we have shown that graphs can have bound states that quantum walks starting on an edge leading into the graph can not access .
we also defined what it means for a quantum walks on a graph to be time - reversal invariant , and used this property to show when transmission amplitudes for particles coming from different directions will be the same .
it should be possible to extend all of the arguments here to the case in which a graph has more than two tails attached to it .
that means that it can have multiple entrances and exits .
this will simply result in the s - matrix of the graph having multiple channels .
we want to solve the equation @xmath92 . substituting in the explicit expression for @xmath175
, we find , by looking at the indicated edges latexmath:[\[\begin{aligned } b_{0b } ) \nonumber \\
b_{0a } \nonumber \\ \nonumber \\
\nonumber \\ our object is to find @xmath88 and @xmath87 from which we can find the transmission and refelection coefficients , respectively , of the graph . keeping this in mind ,
these ten equations can readily be reduced to six @xmath336 these equations can then be solved , giving us @xmath337 note that @xmath88 is just the transmission amplitude , @xmath282 , for the graph , and @xmath87 is its reflection amplitude .
we want to show that eq .
( [ reftrans ] ) is true . a graph is specified by its set of vertices , @xmath338 , and its set of edges , @xmath24 .
a state @xmath339 can be expressed as @xmath340 consider a set of vertices in a graph , @xmath19 , and its complement @xmath341 . defining @xmath342
note that if @xmath343 , then @xmath344 is just the probability that a particle in the state@xmath345 will be found on one of the edges in the set @xmath346 , with similar considerations for @xmath347 , and this justifies the notation .
we can now apply this to the state @xmath351 and choose the set @xmath19 to be the vertices of @xmath123 .
application of @xmath73 to this state gives us @xmath352 and a short calculation shows that @xmath353 .
we also find that @xmath354 inserting these results into eq .
( [ conserve ] ) eq .
( [ reftrans ] ) follows .
we can actually do a bit more than this .
consider eq .
( [ leftright ] ) in which the eigenstate for a particle coming from the left and the eigenstate for a particle coming from the right are given .
the result in the previous paragraph is just @xmath355 and application of eq .
( [ conserve ] ) to the state @xmath356 gives us @xmath357 in addition , if we have the state @xmath358 we can define @xmath359 the fact that @xmath73 is unitary implies that @xmath360 applying this equation to the states @xmath361 and @xmath362 gives us that @xmath363 what all of this has proved , is that the s - matrix for the graph , @xmath364 which relates the incoming amplitudes to the outgoing amplitudes , is a unitary matrix .
99 n. shenvi , j. kempe , and k. b. whaley , phys .
a * 67 * , 052307 ( 2003 ) and quant - ph/0210064 .
a. childs , r. cleve , e. deotto , e. farhi , s. gutman , and d. spielman , proceedings of the 35th acm symposium on the theory of computing ( stoc03 ) ( acm press , new york 2003 ) , pages 59 - 68 , and quant - ph/0209131 .
y. aharanov , l. davidovich , and n. zagury , phys .
rev . a * 48*,1687 ( 1993 ) .
j. watrous , proceedings of the 33rd symposium on the theory of computing ( stoc01 ) ( acm press , new york , 2001 ) , p. 60 .
a. childs , e. farhi , and s. gutmann , quantum information processing * 1 * , 35 ( 2002 ) .
m. hillery , j. bergou and e. feldman , phys .
a * 68 * , 032314 ( 2003 ) .
d. meyer , j. stat .
phys . * 85 * , 551 ( 1996 ) .
a. nayak and a. vishwanath , quant - ph/0010117 ( 2000 ) .
d. aharonov , a. ambainis , j. kempe , and u. vazirani , proceedings of the 33rd symposium on the theory of computing ( stoc01 ) ( acm press , new york , 2001 ) , p. 50 - 59 and quant - ph/0012090 ( 2000 ) .
t. yamasaki , h. kobayashi , and h. imai in c. calude , m. j. dinneen , and f. pepper , editors , unconventional models of computation , third international conference , umc 2002 , kobe , japan , october 15 - 19 , 2002 , proceedings , volume 2509 of _ lecture notes in computer science _
( springer , new york , 2002 ) p. 315 - 330 and
quant - ph/0205045 ( 2002 ) .
e. bach , s. coppersmith , m. paz goldschen , r. joynt , and j. watrous , quant - ph/0207003 . t. d. mackay , s. d. bartlett , l. t. stephenson , and b. c. sanders , j. phys .
a * 35 * , 2745 ( 2002 ) . c. moore and a. russell in j. d. p. rolim and s. vadhan , editors , proceedings random 2002 ( springer , new york , 2002 ) p. 164 - 178 and
quant - ph/0104137 ( 2001 ) .
j. kempe , quant - ph/0205083 ( 2002 ) .
t. a. brun , h. a. carteret , and a. ambainis , phys .
032304 ( 2003 ) and quant - ph/0208195 . v. kendon and b. tregenna , phys .
a * 67 * , 042305 ( 2003 ) and quant - ph/0209005 . j. kempe , contemporary physics * 44 * , 307 ( 2003 ) and quant - ph/0303081 .
b. c. travaglione and g. j. milburn , phys .
a * 65 * , 032310 ( 2002 ) quant - ph/0109076 .
w. dr , r. rausendorf , v. m. kendon , and h. -j .
briegel , phys .
a * 66 * , 052319 ( 2003 ) and quant - ph/0203037 . b. c. sanders , s. d. bartlett , b. treganna , and p. l. knight , quant - ph 0207028 .
z. zhao , j. du , h. li , t. yang , and j. -w .
pan , quant - ph/0212149 .
h. jeong , m. paternostro , and m. s. kim , quant - ph/0305003 .
peter l. knight , eugenio roldan , and j. e. sipe , phys .
a * 68 * , 020301 ( 2003 ) and quant - ph/0304201 .
peter l. knight , eugenio roldan , and j. e. sipe , opt .
* 227 * , 147 ( 2003 ) and quant - ph/0305165 .
d. bouwmeester , i. marzoli , g. p. karman , w. schleich , and j. p. woerdman , phys .
a * 61 * , 013410 ( 2000 ) .
a. g. m. schmidt , b. k. cheng , and m. g. e. da luz , j. phys .
a * 36 * , l545 ( 2003 ) .
e. farhi and s. gutmann , phys .
a * 58 * , 915 ( 1998 ) .
a. zeilinger , h. j. bernstein , d. m. greenberger , m. a. horne , and m. zukowski , controlling entanglement in quantum optics , in _ quantum control and measurement _ , h. ezawa and y. murayama , eds .
( north holland , amsterdam , 1993 ) .
t. kato , _ perturbation theory for linear operators _ ( springer - verlag , new york , 1966 ) , chapter 1 .
|
a number of classical algorithms are based on random walks on graphs .
it is hoped that recently defined quantum walks can serve as the basis for quantum algorithms that will faster than the corresponding classical ones .
we discuss a particular kind of quantum walk on a general graph .
we affix two semi - infinite lines to a general finite graph , which we call tails . on the tails ,
the particle making the walk simply advances one unit at each time step , so that its behavior there is analogous to free propatation .
we are interested in how many steps it will take the particle , starting on one tail and propagating through the graph ( where its propagation is not free ) , to emerge onto the other tail .
the probability to make such a walk in @xmath0 steps and the hitting time for such a walk can be expressed in terms of the transmission amplitude for the graph , which is one element of its s matrix . demonstrating this neccessitates a study of the analyticity properties of the transmission and reflection coefficients of a graph .
we show that a graph can have bound states that can not be accessed by a particle entering the graph from one of the tails .
time - reversal invariance of a quantum walk is defined and used to show that the transmission amplitudes for the particle entering the graph from different directions are the same if the walk is time - reversal invariant .
|
there have been a number of investigations of the force exerted on a charged particle by an aharonov - bohm flux line .
classically , of course , there is no force , so it , like the aharonov - bohm effect itself , is essentially quantum mechanical , vanishing as @xmath2 .
olariu & popescu ( 1983 , 1985 ) show that for certain localized wavepackets ( these are the fast wavepackets described in below ) , the force , along with the momentum it imparts , is negligible unless the centre of the wavepacket hits the flux line .
nielson and hedegrd ( 1995 ) and shelankov ( 2000 ) compute matrix elements of the force operator for stationary states of the same energy .
shelankov ( 1998 ) calculates the transverse force on a stationary incident beam of finite angular width using a paraxial approximation , a result we refer to as _ shelankov s
formula_. this use of the paraxial approximation has been justified by berry ( 1999 ) , who computes the asymptotic deflection of the beam from an exact representation .
peshkin ( 1981 , 1989 ) computes the expectation value of the force when the flux line is enclosed in an impenetrable cylinder .
recent interest in this problem has been stimulated by the analogy with the iordanskii force ( iordanskii 1966 ) exerted on phonons by a vortex in a superfluid ( see , eg , sonin 1975 , sonin 1997 , thouless 1997 , stone 1999 ) , which has been the subject of some recent debate . in this paper
we add to these investigations in several ways .
first , we obtain exact expressions for the lorentz force operator due to an aharonov - bohm flux line and its matrix elements between arbitrary stationary states ; the restriction to on - shell matrix elements , between stationary states of the same energy , agrees with previous results .
we also obtain an exact expression for the time integral of the force , the impulse operator , and compute its expectation values for various kinds of wavepackets .
we show that the force operator can be simply derived , formally at least , from purely kinematic considerations , and give a mathematically rigorous demonstration to justify the results obtained in this way .
the paper is organized as follows . in
, we give a formal derivation of the ( vector ) lorentz force operator due to an aharonov - bohm flux line . pointing out that the nominal magnetic field operator , @xmath3 , is incompatible with gauge invariance ,
we show that that a modification of the canonical commutation relations restores gauge invariance and yields the lorentz force .
an explicit formula for the expectation value for stationary states follows directly . on -
shell matrix elements between stationary states of the same energy are seen to coincide with previous results .
the transverse component of the force is shown to coincide with shelankov s formula in the paraxial limit ( see also shelankov ( 2000 ) ) .
we then compute ( ) the impulse operator , the integral of the force operator in infinite forward and backward time , in the position representation . in
we compute the leading - order expectation value of the impulse for two kinds of wavepackets . for slow wavepackets ( ) , which spread more quickly than they move
, the impulse is inversely proportional to the initial distance between the wavepacket and the flux line , with magnitude and direction depending periodically on the flux parameter @xmath1 . by treating the deflection of a slow wavepacket as a classical scattering
, we obtain an expression for the scattering cross - section which , surprisingly , coincides with the exact result . fast wavepackets ( ) move more quickly than they spread , so much so that the fractional increase in their width as they pass the flux line is small . in this case , the impulse is proportional to the transverse probability density at the flux line , and its transverse component gives a wavepacket analogue of shelankov s formula . in the appendix , we give a rigorous derivation of the force and impulse expectation values for a class of well - behaved wavefunctions . in ,
we compute matrix elements of the force operator for two standard regularizations of the aharonov - bohm flux line , the first where the flux is enclosed in an impenentrable cylinder , and the second where it is distributed uniformly in a cylindrical tube .
the aharonov - bohm results of and are recovered in the limit that the radius of the cylinder goes to zero .
consider a particle of charge @xmath4 and mass @xmath5 moving in the @xmath6-plane in a magnetic field along @xmath7 .
quantum mechanically , the particle is described by the hamiltonian @xmath8 , where @xmath9 is the kinetic momentum and @xmath10 is the vector potential .
the force , ie the rate of change of the kinetic momentum , is given by the appropriately symmetrized lorentz force operator , @xmath11 = \frac{e}{2c}\left ( { { \bi v}}\wedge ( b{\,\hat{{{\bi z } } } } ) - ( b{\,\hat{{{\bi z}}}})\wedge { { \bi v}}\right),\ ] ] where the magnetic field operator is defined by @xmath12.\ ] ] here @xmath13 is the magnetic flux quantum .
the usual commutation relations for position and canonical momentum lead to the usual relation between the magnetic field and the vector potential , namely @xmath14 . however , this relation is incorrect for the vector potential of an aharonov - bohm flux line ( since we are restricting to the plane , we should perhaps say ` flux point ' , but we will follow conventional usage ) . for a flux line at the origin of strength @xmath15 , and in a circularly symmetric gauge ,
the vector potential is given by @xmath16 as is well known , physically meaningful quantities depend only on the fractional part , @xmath17 , of the flux parameter , @xmath1 ; a unit shift in the flux parameter , @xmath18 , is equivalent to the gauge transformation @xmath19 , where @xmath20 as a consequence , a physically observable operator @xmath21 which depends on the flux parameter must transform under @xmath22 according to @xmath23 for example , the kinetic momentum @xmath24 satisfies this relation , as is easily verified . as the magnetic field operator ( [ eq:2 ] ) is expressed in terms of the commutator of components of kinetic momentum , it must also satisfy ( [ eq:6 ] ) .
however , @xmath25 , which is given by @xmath26 , clearly does not satisfy ( [ eq:6 ] ) ; it is invariant under the gauge transformation @xmath22 ( like any operator which depends only on position ) , but is not periodic in @xmath1 .
it follows that substituting @xmath27 for the magnetic field in ( [ eq:1 ] ) can not give the correct expression for the lorentz force operator .
the problem is caused , of course , by the singularity at @xmath28 , and can be avoided by an explicit calculation of the time rate of change of the expectation value of the kinetic momentum for suitably chosen wavefunctions .
this is done in the appendix .
however , we can obtain the same result more easily and more directly from formal arguments .
if the usual canonical commutation relations lead to a magnetic field operator which does not transform correctly under gauge transformations , then the canonical commutation relations must be modified by the flux line . in particular , we show below that , formally , the components of the canonical momentum , @xmath29 and @xmath30 , do not commute equivalently , the partial derivatives @xmath31 and @xmath32 do not commute in the presence of nonzero flux .
first , we note that the partial derivatives @xmath31 and @xmath32 certainly commute when applied to smooth wavefunctions .
that is , @xmath33 \psi = 0\ ] ] for @xmath34 smooth .
applying the gauge transformation @xmath22 to this relation @xmath35 times , we get @xmath36 \left(\rme^{\rmi m\phi}\psi\right ) = 0.\ ] ] the partial derivatives transform according to @xmath37 substituting ( [ eq:10 ] ) into ( [ eq:9 ] ) , and using the differential version of stokes theorem , @xmath38\phi = ( { { \bi \nabla}}\wedge{{\bi \nabla}})_z \phi = 2\pi \delta^2({{\bi r}}),\ ] ] we obtain @xmath39 \left(\rme^{\rmi m\phi } \psi\right ) = 2\pi \delta^2({{\bi r } } ) \rmi m\rme^{\rmi m\phi } \psi = 2\pi \delta^2({{\bi r } } ) \partial_\phi \left(\rme^{\rmi m \phi } \psi \right),\ ] ] where the second equality follows because @xmath40 vanishes at the origin for @xmath34 smooth .
this implies , formally , the operator relation @xmath41 = 2 \pi \delta^2({{\bi r } } ) \partial_\phi,\ ] ] or , equivalently , @xmath42 = -2\pi \rmi \hbar \delta^2({{\bi r } } ) l,\ ] ] where @xmath43 is the canonical angular momentum .
using the modified commutation relation ( [ eq:4 ] ) to evaluate the magnetic field operator ( [ eq:2 ] ) , we get , instead of the classical relation @xmath44 , the result @xmath45 + ( \nabla\wedge { { \bi a}})_z = -\frac{\phi_0}{\hbar } \delta^2({{\bi r}})\lambda,\ ] ] where @xmath46 is the kinetic angular momentum .
it is evident that @xmath47 , and hence @xmath48 , satisfy the transformation law ( [ eq:6 ] ) . throughout this and
the following sections , it will be convenient to represent vectors in the @xmath6-plane as complex scalars . for example , @xmath49 will be represented by @xmath50 .
if @xmath51 is a vector of hermitian operators ( as opposed to real scalars ) , then @xmath52 is the nonhermitian scalar operator whose hermitian and antihermitian parts are @xmath53 and @xmath54 respectively . in this way , the kinetic momentum @xmath24 is represented by the scalar operator @xmath55 it is useful to note that , in general , @xmath56 is represented by @xmath57 .
we proceed to compute the expectation value of the lorentz force . substituting ( [ eq:15 ] ) for the magnetic field and ( [ eq:13 ] ) for the kinetic momentum into the expression ( [ eq:1 ] ) , we get @xmath58\psi(r,\phi)r\,\rmd r , \end{aligned}\ ] ] where we have used @xmath59 . with @xmath60
resolved into its angular momentum components , @xmath61 the integrals in ( [ eq:14 ] ) are trivially evaluated ( note that @xmath62 ) , with the result @xmath63_{r = 0}.\ ] ] like any vector operator , @xmath64 couples only consecutive angular momentum components , @xmath35 and @xmath65 , and , as one would expect , depends only on the behaviour of the wavefunction at the flux line . for well - behaved wavefunctions , the leading - order behaviour of @xmath66 as @xmath67 is given by @xmath68 sufficient conditions for ( [ eq : r=0 ] ) are discussed in the appendix .
here , we note that ( [ eq : r=0 ] ) ensures that the energy density , @xmath69 , is finite at @xmath70 . substituting ( [ eq : r=0 ] ) into ( [ eq : f2 ] ) , we get @xmath71 the terms in the sum ( [ eq : f2two ] ) vanish unless @xmath72 , where @xmath73 denotes the integer part of the flux parameter .
thus , only the ` pinioned ' components of the wavefunction , @xmath74 and @xmath75 those whose angular momentum quantum numbers are nearest the flux parameter @xmath1 contribute to the force expectation value ( [ eq : f2 ] ) . from ( [ eq : f2two ] ) , @xmath76 or , equivalently , @xmath77 where the state @xmath78 corresponds to the singular wavefunction @xmath79 so that @xmath80 .
it is readily verified that the force operator ( [ eq : forceop ] ) transforms according to ( [ eq:6 ] ) under the gauge transformation ( [ eq:5 ] ) .
a rigorous derivation of the expectation value ( [ eq : fexpect ] ) for suitably chosen wavefunctions is given in the appendix . in the preceding derivation
, the force operator due to a flux line , like the force for a nonsingular potential , is derived from kinematics , specifically from the commutation relations .
the derivation does not require the solution of the schrdinger equation .
thus , it is straightforward to generalize to the case of more than one flux line ( for which solutions of the schrdinger equation are , in general , not available ) ; the force operator is just a sum of contributions ( [ eq : forceop ] ) centred around each flux line .
however , to calculate the force on stationary states , or the time dependence of the force on nonstationary states , it is necessary to solve the schrdinger equation .
as is well known , eigenstates of the aharonov - bohm hamiltonian with energy @xmath81 and angular momentum @xmath82 are given by @xmath83 where @xmath84 is a bessel function . from the small-@xmath85 behaviour , @xmath86 , and the reflection formula , @xmath87
, we obtain from ( [ eq : forceop ] ) the matrix elements @xmath88 for @xmath89 , ie for stationary states with the same energy , we obtain the on - shell matrix elements @xmath90 in agreement with results of nielson and hedegrd ( 1995 ) and shelankov ( 2000 ) .
a general stationary state @xmath91 is a superposition of eigenstates with @xmath0 fixed , and may be taken to be of the form @xmath92 for @xmath93 , @xmath94 corresponds to a scattered plane wave incident from the right ( aharonov & bohm 1959 ) . from ( [ eq : onshellmatrix ] ) , we get the expectation value @xmath95 shelankov ( 1998 ) has obtained an approximate formula for the transverse component of the force acting on a stationary beam of finite angular width .
his analysis is carried out in a singular gauge , in which the vector potential vanishes everywhere except along the @xmath96-axis . a stationary beam incident from the right ( in fact
, shelankov takes the beam incident from the left , but we revert to the convention of aharonov and bohm ) is taken to be of the form @xmath97 . treating @xmath98 as small
compared to @xmath99 amounts to a paraxial approximation , in which the wave evolves freely in @xmath100 ( with @xmath100 playing the role of time ) for @xmath101 , and is scattered by the vector potential at @xmath102 .
the change @xmath103 in the transverse kinematic momentum , @xmath104 , is then calculated to be @xmath105 where @xmath106 is the incident wave at @xmath107 , expressed here in terms of its transverse fourier amplitudes @xmath108 .
multiplying @xmath103 by the incident flux , which is given paraxially by @xmath109 , gives _
shelankov s formula _ for the transverse force , @xmath110
we now show that the @xmath96-component of the exact force expectation value , ie the imaginary part of ( [ eq : stationary ] ) , coincides with shelankov s formula ( [ eq : shelforce ] ) in the paraxial regime .
( shelankov ( 2000 ) gives the same argument . )
as discussed by berry ( 1999 ) , the state ( [ eq : super ] ) can alternatively be viewed as a superposition of scattered waves incident from the directions @xmath111 , with amplitudes @xmath112 related to the coefficients @xmath113 according to @xmath114 the paraxial approximation is valid for @xmath112 strongly peaked around @xmath115 , with angular width @xmath116 . in this case , berry ( 1999 ) has shown that @xmath117 . from ( [ eq : psiin ] ) and ( [ eq : b_m ] ) , it then follows that @xmath118 for @xmath119 , so that @xmath120 .
for nonstationary wavepackets @xmath34 , whose wavefunctions are not eigenfunctions of the hamiltonian , the expectation value of the force does not itself have much physical significance .
it depends on the behaviour of the wavefunction near the flux line , regardless of where the wavepacket is localized , and can oscillate rapidly as the wavepacket evolves . of greater physical interest
is the impulse imparted to the particle over the course of its evolution , either in the past or future .
let @xmath121 denote the time - evolved force and kinetic momentum operators
. then @xmath122 .
the forward ( + ) and backward ( - ) impulse operators are defined by @xmath123 let us compute the kernal of the impulse operator in the position representation , @xmath124 . from the completeness relation ,
@xmath125 we obtain @xmath126 from the expression ( [ eq : matrix ] ) for the matrix elements @xmath127 , the only contribution to the double sum in ( [ eq : res2a ] ) is from the term @xmath128 , @xmath129 . substituting the eigenfunctions ( [ eq : eigen ] ) , and letting @xmath130
denote the polar coordinates of @xmath131 , we get @xmath132 the @xmath0- and @xmath133-integrals are of the form ( abramowitz & stegun 1970 ) @xmath134 substituting this result into ( [ eq : res2 ] ) , we get @xmath135 with the substitution @xmath136 , the remaining integral in ( [ eq : i3 ] ) is of the elementary form @xmath137 with @xmath138 . substituting ( [ eq : t - int ] ) into ( [ eq : i3 ] )
, we obtain @xmath139 one can verify that this expression transforms correctly , ie according to ( [ eq:6 ] ) , under the gauge transformation ( [ eq:5 ] ) .
the denominator in ( [ eq : ifinal ] ) can be alternatively expressed as @xmath140 here @xmath141 , the derivative of the principal part , acts on functions @xmath142 according to @xmath143 where @xmath144 denotes the odd part of @xmath145 . for subsequent calculations ,
however , the integral representation ( [ eq : t - int ] ) will be more convenient .
we parameterize wavepackets by their position @xmath146 , width @xmath147 and kinetic momentum @xmath148 .
a convenient form is @xmath149 here @xmath150 is a smooth normalized function localized at the origin with unit width and vanishing average ( dimensionless ) momentum , ie @xmath151 we assume that @xmath152 , so that the wavepacket @xmath34 is localized far from the flux line .
the phase factor @xmath153 in ( [ eq : psi ] ) insures that @xmath154 is the kinetic , rather than the canonical , momentum of the wavepacket ; its branch is chosen so that the phase factor is continuous over the region where @xmath34 is appreciable . from ( [ eq : ifinal ] ) and ( [ eq : psi ] ) , the expectation value of the impulse is given by @xmath155 since @xmath156 has unit width , the integrand in ( [ eq : iexpect2a ] ) is appreciable only for @xmath157 and @xmath158 . in this region , we can , to leading order in @xmath159 , replace the phase factors @xmath160 and @xmath161 by @xmath162 and @xmath163 , respectively , where @xmath164 is the polar angle of @xmath146 .
likewise , we can replace the factor @xmath165 by @xmath166 . with the change of variables @xmath167 and @xmath168 and the integral representation ( [ eq : t - int ] ) , ( [ eq : iexpect2a ] ) becomes @xmath169 thus , the direction of the impulse , @xmath170 , is given by @xmath171 for @xmath172 , the forward impulse is directed away from the flux line , and the backwards impulse towards the flux line .
there are two parameter regimes where the expression ( [ eq : k2 ] ) has a simple asymptotic form , namely @xmath173 , which corresponds to slow wavepackets , and @xmath174 , which corresponds to fast wavepackets .
these cases are discussed separately below .
the condition @xmath173 implies that the wavepacket spreads ( with velocity @xmath175 ) more quickly than it moves ( with velocity @xmath176 ) .
since @xmath156 has unit width , the integrand in ( [ eq : k2 ] ) is appreciable only for @xmath177 of order one . in this case
, for @xmath173 , the phase factor @xmath178 is nearly equal to one . on the other hand , the phase factor @xmath179 oscillates rapidly in this region , and
hence renders the integral small , unless @xmath180 is small , of order @xmath159 . for @xmath180 of order
@xmath159 , the phase factor @xmath181 is nearly equal to one for @xmath177 of order one . to
leading order in @xmath159 and @xmath182 , ( [ eq : k2 ] ) becomes @xmath183 where @xmath184 denotes the normalized fourier transform of @xmath150 .
letting @xmath185 denote the probability distribution for the direction , @xmath186 , of the dimensionless momentum , @xmath187 , we can write @xmath188 note that if @xmath150 is circularly symmetric , then @xmath189 is equal to @xmath190 . to leading order in @xmath159 and @xmath182 ,
the impulse ( [ eq : slow ] ) is independent of the width and momentum of the wavepacket , and is of order @xmath191 ( ie , inversely proportional to the distance from the flux line ) .
this is much smaller than the dispersion of the momentum , which is of order @xmath192 .
therefore , to detect the impulse on slow wavepackets experimentally , one would have to perform a large number of measurements ( on the order of @xmath193 ) of the asymptotic momentum on an ensemble of identically prepared systems . by treating the motion of the centre of a slow wavepacket as a classical trajectory
, we can derive an expression for the scattering cross - section @xmath194 using the classical formula , @xmath195 here @xmath196 is the impact parameter , and @xmath197 is the angular direction of the outgoing trajectory . consider a slow wavepacket ( [ eq : psi ] ) centred on the @xmath96-axis at @xmath198 at @xmath199 ( thus , @xmath200 and @xmath201 ) , moving in the @xmath202 direction with kinetic momentum @xmath154 .
for simplicity , we take @xmath150 to be circularly symmetric , so that @xmath203 .
let @xmath204 denote the angle between the velocities at @xmath205 and @xmath206 , and @xmath207 the angle between the velocities at @xmath208 and @xmath209 ( see fig 1 ) .
from ( [ eq : slow ] ) , these are given by @xmath210 ( thus @xmath211 ) .
[ fig1 ] because the impulse is circularly symmetric , the angles @xmath212 are unchanged if we rotate the entire system so that the incoming velocity , at @xmath213 , is in the @xmath202 direction . in this case , the direction of the outgoing beam is given by @xmath214 to determine the impact parameter @xmath196 , we appeal to classical angular momentum conservation , @xmath215 , where @xmath216 is the speed at @xmath213 , and @xmath217 and @xmath218 are the distance and speed at the point of closest approach to the flux line . for the aharonov - bohm hamiltonian ( and , indeed , for any purely magnetic hamiltonian ) , the speed @xmath219 is a conserved quantity .
thus @xmath220 .
we take the point of closest approach to occur at @xmath206 ( when the velocity of the wavepacket is orthogonal to its position ) , so that @xmath221
. then , from ( [ eq : classicalxsec ] ) , ( [ eq : betapm ] ) and ( [ eq : theta= ] ) , @xmath222 surprisingly , the expression ( [ eq : xsection ] ) agrees with the exact result found by aharonov & bohm ( 1959 ) . of course
, the preceding should not be regarded as a legitimate derivation of the scattering cross - section .
apart from certain _ ad hoc _ elements ( eg , circularly symmetric wavepacket and the determination of the impact parameter ) , a proper derivation of the cross - section from time - dependent solutions of the schrdinger equation requires wavepackets ( unlike the slow ones used here ) whose momentum is sharp .
still , it is interesting to ask whether or not this agreement is purely fortuitous .
a wavepacket initially at a distance @xmath166 from the flux line with kinetic momentum @xmath154 reaches its point of closest approach to the flux line after a time @xmath223 of order @xmath224 .
it spreads at a speed @xmath225 of order @xmath226 .
thus , at closest approach it will have spread a distance of order @xmath227 . for this to be much less than the width @xmath147
, we require @xmath228 , which is just the condition for fast wavepackets . let @xmath229 denote the @xmath230-integral which appears in the impulse expectation value ( [ eq : k2 ] ) . because @xmath156 is of unit width , the integrand is appreciable only for @xmath177 of order @xmath231 . for @xmath174 ,
the phase factor @xmath232 in ( [ eq : k2 ] ) is rapidly oscillating , and hence leads to a vanishingly small integral unless it is balanced by the phase factor @xmath233 .
for such a balancing to take place , @xmath180 must be large , of order @xmath234 .
therefore , the quadratic phase factor @xmath235 is rapidly oscillating , so ( [ eq : uvecint ] ) can be evaluated using the stationary phase approximation . to leading order in @xmath236 , we obtain @xmath237 substituting into ( [ eq : k2 ] )
, we get @xmath238 where we have used ( [ eq : psi ] ) to express the integral in terms of the wavefunction @xmath34 .
note that @xmath34 behaves for small @xmath239 as @xmath240 or @xmath241 ( cf ( [ eq : r=0 ] ) ) , so that the integral in ( [ eq : kfast ] ) is convergent . in
what follows , let us assume for concreteness that @xmath242 is directed along -@xmath243 , so that the wavepacket is moving to the left .
we write @xmath244 .
unless the wavepacket is centred near the @xmath100-axis ( specifically , unless @xmath245 ) , @xmath246 will be negligible over the range of integration in ( [ eq : kfast ] ) .
thus , to leading order in @xmath159 , we may take @xmath247 . substituting this result into ( [ eq : kfast ] )
, we obtain the expression @xmath248 for the impulse .
since the wavepacket is centred near @xmath249 , the @xmath100-integral in ( [ eq : ifastaa ] ) is negligible unless @xmath250 in the forward @xmath251 case ( so that the wavepacket is moving towards the flux line ) , or unless @xmath252 in the backward ( - ) case ( so that the wavepacket is moving away from the flux line ) . assuming that @xmath253 , the main contribution to the integral comes from @xmath254 , so that , to leading order in @xmath159 , we can replace the factor @xmath255 by @xmath256 in ( [ eq : ifastaa ] ) , and extend the lower limit of the @xmath100-integral to @xmath257 . letting @xmath258 denote the wavepacket s probability density in @xmath96 ( the direction transverse to the incident velocity ) , we obtain , to leading order in @xmath159 and @xmath259 , the expression @xmath260 for the impulse on fast wavepackets . here
@xmath261 is the unit step function .
the impulse ( [ eq : ifast ] ) is independent of the wavenumber @xmath0 . to leading order
, it vanishes for wavepackets which miss the flux line ( eg , @xmath262 , or @xmath253 ) , as shown previously by olariu & popescu ( 1983 , 1985 ) . for fast wavepackets which hit the flux line , taking @xmath263 to be of order @xmath264 for @xmath265
, we get that the impulse is of order @xmath192 . therefore , it is of the same order as the dispersion in momentum , in contrast with the case of slow wavepackets , for which the impulse is much smaller ( by a factor of @xmath159 ) than the dispersion .
the @xmath96-component of the forward impulse , ie the imaginary part of ( [ eq : ifast ] ) , is given in the forward case by @xmath266 this can be regarded as an analogue in the time domain of shelankov s formula ( [ eq : shelimpulse ] ) for the transverse momentum imparted to a stationary paraxial beam .
two well - known regularizations of the aharonov - bohm flux line are to enclose the flux in an impenetrable cylindrical barrier , or to distribute the flux uniformly in a cylindrical tube . here
we show that the force and impulse operators in both cases approach the aharonov - bohm limit , in a sense to be explained , as the radius @xmath267 of the cylinder approaches zero . in a circularly symmetric gauge ,
the vector potential for both models is of the form @xmath268 .
the kinetic momentum is given by @xmath269 and the regularized hamiltonian by @xmath270 the eigenfunctions of the hamiltonian and kinetic angular momentum , with energy @xmath271 and kinetic angular momentum @xmath82 , are of the form @xmath272 the radial eigenfunctions @xmath273 are taken to be real and normalized , like the aharonov - bohm radial eigenfunctions @xmath274 , according to @xmath275 these conditions determine the radial eigenfunctions up to an overall sign , which is fixed by requiring that @xmath276 approach @xmath277 as @xmath267 approaches zero .
let @xmath278 denote a smooth , normalized probability amplitude for energy and angular momentum .
let @xmath34 and @xmath279 denote the corresponding coordinate wavefunctions for the aharonov - bohm and regularized hamiltonians , respectively .
that is , @xmath280 from the preceding discussion , it is clear that @xmath279 approaches @xmath34 pointwise as @xmath267 approaches zero .
it turns out that their force and impulse expectation values also coincide as @xmath281 , ie @xmath282 @xmath283 note that ( [ eq : forcelimit ] ) and ( [ eq : impulselimit ] ) do not imply , nor is it the case , that the operators @xmath284 and @xmath285 approach their aharonov - bohm counterparts , @xmath64 and @xmath286 , as @xmath267 approaches 0 . indeed ,
neither does the regularized hamiltonian @xmath287 approach the aharonov - bohm hamiltonian @xmath288 ; given @xmath289 , one can construct wavefunctions whose energy expectation values with respect to @xmath287 and @xmath288 differ by arbitrarily large amounts . instead of ( [ eq : forcelimit ] ) and ( [ eq : impulselimit ] ) , we show below , for the enclosed and distributed fluxes separately , that the eigenstate matrix elements of the regularized force operator approach the aharonov - bohm limit as @xmath281 , ie , @xmath290 formally , of course , ( [ eq : forcelimit ] ) and ( [ eq : matrixeps ] ) are equivalent . however , for the sake of brevity we shall omit the details required for a rigorous demonstration .
these details are straightforward to supply , and are similar to those given in the first part of the appendix .
the result ( [ eq : impulselimit ] ) for the impulse follows from the corresponding result ( [ eq : forcelimit ] ) for the force , once it has been established that the force expectation values @xmath291 and @xmath292 are integrable in time . in the appendix
it is shown that , in the aharonov - bohm case , the force expectation value decays as @xmath293 ; a similar argument may be given for the regularized force .
the kinetic momentum @xmath294 and hamiltonian @xmath287 have the same operational form as in the aharonov - bohm case , but act on wavefunctions defined on @xmath295 which vanish at @xmath296 . for two such wavefunctions , @xmath279 and @xmath297 , assumed to be smooth and normalized , we have @xmath298 \nonumber\\ & = & \frac{i}{\hbar}\big ( { \left\langle h^\epsilon \psi^\epsilon \right.\left|m{{\cal v}}^\epsilon \eta^\epsilon\right\rangle } -{\left\langle \psi^\epsilon \right.\left|h^\epsilon\left(m{{\cal v}}^\epsilon \eta^\epsilon\right)\right\rangle } \big).\end{aligned}\ ] ] the last equality follows from the fact that the commutator @xmath299 $ ] is proportional to the lorentz force operator ( [ eq:1 ] ) , which vanishes for the enclosed flux .
however , the final expression in ( [ eq : fenclosed ] ) does not vanish ; the relation @xmath300 , where @xmath301 need not hold , because @xmath302 need not vanish at @xmath303 ( alternatively , @xmath304 is not in the domain of @xmath287 ) .
indeed , integration by parts in ( [ eq : ha ] ) gives @xmath305 r\,\rmd r\,\rmd \phi\nonumber\\ & = & -\frac{\hbar^2}{2m}\int_0^{2\pi } \psi^{\epsilon*}_r(\epsilon,\phi)\xi^\epsilon(\epsilon,\phi)\,\rmd \phi\end{aligned}\ ] ] from ( [ eq : km2 ] ) and ( [ eq : xi(r ) ] ) , @xmath306 substituting this result into ( [ eq : boundaryterm ] ) , we get @xmath307 a result obtained previously by peshkin ( 1981 , 1989 ) . note that if we were to substitute , for @xmath308 and @xmath309 , the leading - order behaviour ( [ eq : r=0 ] ) of aharonov - bohm wavefunctions , we would recover , formally , the aharonov - bohm result ( [ eq : fexpect ] ) for the force expectation value .
instead , we take @xmath279 and @xmath297 in ( [ eq : fa2 ] ) to be eigenfunctions of the regularized hamiltonian .
then @xmath310 ( strictly speaking , this is not legitimate , as there would appear boundary terms at @xmath311 in ( [ eq : boundaryterm ] ) .
however , these would vanish when we consider expectation values , as in ( [ eq : forcelimit ] ) . ) to evaluate ( [ eq : fa3 ] ) we need the derivatives of the radial eigenfunctions at @xmath303 .
the radial wavefunctions themselves are given by @xmath312 where @xmath313 is the neumann function .
the constant @xmath314 is determined by the normalization condition ( [ eq : norm ] ) , and is given by @xmath315 and , to leading order in @xmath267 , by @xmath316 the wronskian relation , @xmath317 , implies that @xmath318 substituting ( [ eq : ceps ] ) and ( [ eq : r ] ) into ( [ eq : fa3 ] ) , we obtain , to leading order in @xmath267 , @xmath319 in the limit @xmath281 , only the @xmath72 term survives , and the reflection formula for the @xmath320-function gives @xmath321 in accord with ( [ eq : matrixeps ] ) . the distributed flux model was used by nielsen and hedegrd ( 1995 ) to obtain , from the force balance equations , the on - shell matrix elements of the force in the limit @xmath281 .
here we carry out a different calculation to obtain the general matrix elements of the force .
it suffices to consider the case @xmath322 ( the case of negative flux is obtained from time - reversal ) .
the vector potential is given by @xmath323 corresponding to the magnetic field @xmath324 , where @xmath261 is the unit step function .
in this case , the force operator is just the lorentz force ( [ eq:1 ] ) .
it is convenient to introduce the dimensionless radial coordinate @xmath325 , so that the interior of the flux tube is given by @xmath326 .
the kinetic momentum is given by @xmath327 then @xmath328.\end{aligned}\ ] ] the matrix elements of the force are given by @xmath329,\end{aligned}\ ] ] where @xmath330 denotes the radial eigenfunction expressed in terms of the scaled variable @xmath177 , and @xmath331 inside the flux tube , the radial eigenfunctions are given by ( landau & lifschitz 1965 ) @xmath332 where @xmath333 is the confluent hypergeometric function ( abramowitz & stegun 1970 ) . outside the flux tube , @xmath334 the coefficients @xmath314 , @xmath335 and @xmath336 are determined by requiring the radial eigenfunction and its first derivative to be continuous at @xmath296 ( the second derivative is then continuous there as well , as it turns out ) , and by the normalization condition @xmath337 which follows from ( [ eq : norm ] ) .
to evaluate the force matrix element ( [ eq : fdist ] ) , we only require the function inside the flux cylinder .
straightforward algebra gives the coefficient @xmath314 , to leading order in @xmath267 , as @xmath338},\ ] ] where @xmath339 and @xmath340 and @xmath341 denote the values of @xmath342 and @xmath343 at @xmath344 . substituting ( [ eq : teps ] ) and ( [ eq : cepsapprox ] ) into ( [ eq : fdist ] ) , we find that that @xmath345 is of order @xmath346 , and therefore vanishes in the limit @xmath281 unless @xmath72 .
we obtain @xmath347 where @xmath348 and @xmath349 as we show below , @xmath350 , or , equivalently , @xmath351 with this identity , ( [ eq : fdistpen ] ) gives the required result ( [ eq : matrixeps ] ) . to establish the identity ( [ eq : identity ] ) , it is convenient to express @xmath352 in terms of @xmath353 by means of the recurrence relation ( abramowitz & stegun 1970 ) @xmath354 which implies that @xmath355 with the differential equation @xmath356 it is straightforward to show that the integrand on the left - hand side of ( [ eq : identity ] ) is given by @xmath357 , where @xmath358 @xmath359 vanishes , whereas @xmath360 , with the aid of ( [ eq : frecur ] ) and ( [ eq : diffeq ] ) , is seen to be equal to the right - hand side of ( [ eq : identity ] ) . we thank professor sir mv berry and professor d khemelnitskii for stimulating discussions .
the force and impulse due to an aharonov - bohm flux line can be calculated rigorously for suitably well - behaved wavefunctions @xmath34 .
we will take these to be such that @xmath361 using standard arguments , one can show that ( [ eq : prop0 ] ) implies the following properties of @xmath34 and @xmath362 , where @xmath288 is the aharonov - bohm hamiltonian : @xmath363 and @xmath364 where , in general , @xmath365 ( in fact , properties ( [ eq : properties ] ) and ( [ eq : prop2 ] ) are also shared by @xmath366 , for @xmath367 . the argument to follow would hold under weaker conditions , but we assume ( [ eq : prop0 ] ) for simplicity . ) the expectation value of force is given by @xmath368 where @xmath369 is given by ( [ eq:13 ] ) . from ( [ eq : properties ] ) and ( [ eq : prop2 ] ) , it is evident that the @xmath370-integral in ( [ eq : frig1 ] ) converges absolutely .
this allows us to introduce a factor @xmath371 in the integrand , and then take the limit of the integral as @xmath372 .
this gaussian factor will justify subsequent reorderings of operations .
note that the integral can not be expressed in terms of the expectation value of the commutator @xmath373 $ ] , because of the singularity in the radial derivative of @xmath34 at the origin ( specifically , @xmath374 is not in the domain of @xmath288 ) .
we introduce the eigenfunction expansion @xmath375 and a similar expansion for @xmath362 , with @xmath278 replaced by @xmath376 . using standard arguments
, one can show that ( [ eq : prop0 ] ) implies that the differential operator @xmath369 , when applied to @xmath60 and @xmath377 , can be taken inside the @xmath35-sum and @xmath0-integral .
the recurrence relation , @xmath378 implies that @xmath379 substituting ( [ eq : psiexpand ] ) and ( [ eq : mvbessel ] ) into ( [ eq : frig1 ] ) , along with the eigenfunction expansion of @xmath380 with coefficients @xmath381 , we obtain @xmath382 the sums and integrals in ( [ eq : frig2 ] ) are uniformly and absolutely convergent , and can be interchanged . on performing the @xmath383-integral , the sum on @xmath384 collapses to the single term @xmath385 .
we obtain @xmath386 where , for @xmath387 , @xmath388 and , for @xmath72 , @xmath389 below , in [ secmnea ] , we show that the contributions from the @xmath387 terms vanish in the limit , ie @xmath390 while , in [ secmea ] , we show for the @xmath72 term that @xmath391 substitution of ( [ eq : mnea ] ) and ( [ eq : mea ] ) into ( [ eq : frig3 ] ) gives @xmath392 this is equivalent to the matrix element ( [ eq : matrix ] ) , obtained formally in section [ sec : force - operator ] .
we note that , with @xmath393 , the integrals in ( [ eq : keps2mnea ] ) and ( [ eq : keps2mea ] ) correspond to singular ( ie , not absolutely convergent ) cases of the discontinuous weber - schafheitlin integral ( abramowitz & stegun 1970 ) .
formal evaluation of these integrals would give ( [ eq : mnea ] ) and ( [ eq : mea ] ) immediately .
the arguments in [ secmnea ] and [ secmea ] serve to justify these formal results . to obtain the force expectation value ( [ eq : fexpect ] ) , we express @xmath394 and @xmath395 in ( [ eq : frig4 ] ) in terms of @xmath34 to obtain @xmath396 note that the convergence factor @xmath397 can be introduced , and the limit @xmath281 taken outside the integral , since , by ( [ eq : prop0 ] ) , the @xmath0- and @xmath133-integrals in ( [ eq : frig4 ] ) are absolutely convergent . by ( [ eq : properties ] ) and ( [ eq : prop2 ] ) ,
the @xmath239- and @xmath398-integrals in ( [ eq : frig5 ] ) are absolutely convergent , so that , for @xmath399 , we can interchange the order of integration .
the @xmath0- and @xmath133-integrals can be evaluated using ( [ eq : integral ] ) , with the result @xmath400 where the coefficients @xmath401 and @xmath402 are given in ( [ eq : prop2 ] ) .
this is just the result ( [ eq : fexpect ] ) of section [ sec : force - operator ] .
concerning the impulse , it is straightforward to justify , using arguments like the preceding ones , the calculations of section [ sec : impulse - operator ] leading to ( [ eq : iexpect2a ] ) .
it is only necessary to check that the time - dependent expectation value , @xmath403 , is integrable in @xmath404 .
@xmath403 is given by an expression like ( [ eq : frig4 ] ) , but with @xmath394 and @xmath395 modulated by the factors @xmath405 and @xmath406 respectively .
we have that @xmath407 for large @xmath404 ( the integration by parts is justified by ( [ eq : properties ] ) and ( [ eq : prop2 ] ) ) .
similarly , @xmath408 .
it follows that @xmath403 falls off as @xmath293 .
given functions @xmath156 and @xmath409 defined on a domain @xmath410 , we will say that @xmath156 is dominated by @xmath409 if , for some constant @xmath411 , @xmath412 throughout @xmath410 . for functions indexed by an integer @xmath35 , eg @xmath340 and @xmath413 , we will say that @xmath340 is dominated by @xmath413 if @xmath414 for some @xmath411 which does not depend on @xmath35 .
thus , from ( [ eq : prop0 ] ) , @xmath415 for large argument , it follows that @xmath417 is dominated by @xmath419 for @xmath85 real .
therefore , the left - hand side of ( [ eq : mnea ] ) is dominated by @xmath420 let us divide the domain of the @xmath421-integral into regions inside and outside the strip @xmath422 . inside ,
the integrand is dominated by @xmath423 ; thus the integral over the strip is dominated by @xmath424 . outside the strip , the integrand is dominated by @xmath425 ; thus the @xmath421-integral outside the strip is dominated by @xmath426 .
therefore , the expression in ( [ eq : dominus ] ) is dominated by @xmath427 which vanishes as @xmath372 . substituting the series expansion @xmath428 and a similar expansion for @xmath429
, we get that @xmath430 for @xmath289 , the @xmath239-integral and @xmath431 and @xmath432-sums are absolutely convergent for all @xmath433 . inserting in ( [ eq : besselint ] ) the integral representation for the reciprocal of the beta - function ( gradshteyn and ryzhik 1980 ) @xmath434 we may perform the sums to obtain @xmath435 substituting this result into ( [ eq : keps2mea ] ) , we get @xmath436 where the exponent @xmath437 is given by @xmath438 it is clear that the main contribution to the @xmath223-integral in ( [ eq : kepsm = a ] ) comes from the neighbourhood of @xmath439 . if @xmath437 is expanded about @xmath439 to second order , the @xmath223-integral yields an error function , whose leading - order asymptotics as @xmath281 leads directly to the required result ( [ eq : mea ] ) .
however , the next term in the asymptotic expansion is not uniformly bounded in @xmath0 and @xmath133 it contains a factor @xmath440 so we must take some additional care .
to proceed , we divide the domain of the @xmath441-integral into the three regions specified below , writing the left hand side of ( [ eq : mea ] ) as @xmath442 and analyzing the contribution from each region separately .
let @xmath443 denote the region @xmath444 , @xmath445 , where @xmath446 is chosen to satisfy @xmath447 in this region , the exponential factor in ( [ eq : kepsm = a ] ) is bounded , and @xmath448 is dominated by @xmath449 . the coefficients @xmath450 and @xmath451 are bounded , and the area of @xmath443 is dominated by @xmath452 , so that the contribution from @xmath443 in ( [ eq:3parts ] ) is dominated by @xmath453 . given the choice of @xmath446 , this vanishes as @xmath281 .
let @xmath454 be the region @xmath444 ,
@xmath455 and @xmath456 , where @xmath457 is chosen to satisfy @xmath458 ( since @xmath459 , the inequality ( [ eq : gammaineq ] ) can be always be satisfied ) . since @xmath460 for @xmath461 , the factor @xmath462 is dominated by the gaussian @xmath463 , where @xmath464 .
thus @xmath465 is dominated by @xmath466 .
then , from ( [ eq : kepsm = a ] ) , @xmath467 is dominated by @xmath468 in @xmath454 . from ( [ eq : prop0 ] )
, @xmath469 is integrable over the region @xmath444 .
therefore , the contribution from @xmath454 to ( [ eq:3parts ] ) is dominated by @xmath470 ( the additional factor of @xmath471 is due to the fact that the integral is confined to @xmath472 . given the choice of @xmath457 , this vanishes as @xmath281 .
the remaining region @xmath473 is given by @xmath444 , @xmath455 and @xmath474 .
integrating by parts with respect to @xmath223 in ( [ eq : kepsm = a ] ) , we get that @xmath475 the first term gives @xmath476 its contribution to the integral over @xmath473 in ( [ eq:3parts ] ) yields , in the limit @xmath281 , the required result ( [ eq : mea ] ) .
it remains to show that the contribution to the @xmath473-integral from the remaining terms in ( [ eq : byparts ] ) vanishes in the limit @xmath477 , it is readily seen that the contribution from the second term vanishes exponentially with @xmath267 .
for the third term , we note that , on the interval @xmath478 , @xmath479 is dominated by @xmath480 , and @xmath462 is dominated by @xmath481 .
therefore , the integral @xmath482 is dominated by @xmath483 .
thus , the third term in ( [ eq : byparts ] ) is dominated by @xmath484 , which on @xmath473 is dominated by @xmath485 .
the contribution to the integral over @xmath473 in ( [ eq:3parts ] ) is dominated by @xmath486 , which , by the choice of @xmath457 , vanishes as @xmath281 .
|
we calculate the force operator for a charged particle in the field of an aharonov - bohm flux line .
formally this is the lorentz force , with the magnetic field operator modified to include quantum corrections due to anomolous commutation relations . for stationary states ,
the magnitude of the force is proportional to the product of the wavenumber @xmath0 with the amplitudes of the ` pinioned ' components , the two angular momentum components whose azimuthal quantum numbers are closest to the flux parameter @xmath1 .
the direction of the force depends on the relative phase of the pinioned components . for paraxial beams ,
the transverse component of our expression gives an exact version of shelankov s formula [ shelankov a 1998 _ europhys .
lett . _ * 43 * , 623 8 ] , while the longitudinal component gives the force along the beam .
nonstationary states are treated by integrating the force operator in time to obtain the impulse operator .
expectation values of the impulse are calculated for two kinds of wavepackets . for slow wavepackets , which spread faster than they move
, the impulse is inversely proportional to the distance from the flux line . for fast wavepackets , which spread only negligibly before their closest approach to the flux line ,
the impulse is proportional to the probability density transverse to the incident direction evaluated at the flux line . in this case
, the transverse component of the impulse gives a wavepacket analogue of shelankov s formula .
the direction of the impulse for both kinds of wavepackets is flux dependent .
we give two derivations of the force and impulse operators , the first a simple derivation based on formal arguments , and the second a rigorous calculation of wavepacket expectation values .
we also show that the same expressions for the force and impulse are obtained if the flux line is enclosed in an impenetrable cylinder , or distributed uniformly over a flux cylinder , in the limit that the radius of the cylinder goes to zero .
[ aharonov - bohm force and impulse ]
|
Subsets and Splits
Data Mining Abstracts
This query searches for and retrieves 2000 documents containing the phrase 'data mining' in the abstract, providing limited but specific document samples.